The Textual Composition of the “Practices of Secret Mantra Approach” in Jñānakīrti’s Tattvāvātara

Abstract

Jñānakīrti (Tib. Ye shes grags pa), an eminent monk of the late Indian Buddhist period, composed the Tattvāvātara (De kho na nyid la ’jug pa, Realizing Reality), of which only a Tibetan translation exists in the Tibetan Tripiṭaka-Tanjur. The treatise is considered an exposition of the Mahāmudrā teachings, with the chapter entitled “Practices of the Secret Mantra Approach” (gSang sngags kyi sgo’i spyod pa) forms a large part. However, this part has been less frequently discussed. This chapter guides the practice of Mahāmudrā non-dual yoga, which is intended for practitioners with superior faculties. The core content of this chapter can be subsumed under the following two aspects: Mahāmudrā teachings involve practicing insight (prajñā), which represents the theory of meditation, i.e., the idea of emptiness (śūnyatā); it also involves practicing skillful means (upāya), which includes the methods of cultivation, such as tantric rituals such as Vajradhātu maṇḍala visualization. From the perspective of compositional length, the first half of the text contains numerous quotations of verses, with several being from Nāgārjuna’s Mūlamadhyamakakārikā, while the second half mainly draws references from the Tattvasaṃgrahatantra and the Guhyasamājatantra. More attention should be paid to the juxtaposition of the Mahāmudrā teachings with the Tattvasaṃgrahatantra and the Guhyasamājatantra, which reflect the early form of the Mahāmudrā teachings as they were introduced into Tibetan Buddhism.

1. Introduction

In the intellectual crucible of late Indian Buddhism, Jñānakīrti’s Tattvāvātara (De kho na nyid la ’jug pa, Realizing Reality, “TA” hereafter) emerged as a distinctive bridge connecting Madhyamaka philosophy with Tantric practices. Jñānakīrti’s work offers a rare historical cross-section: in its early days of introduction to Tibet, Mahāmudrā (Great Seal) was not an independent sectarian system but rather a Tantric-influenced Madhyamaka path of spiritual cultivation. This study takes this text, commonly regarded as an edge feature, as the entry point to demonstrate that what it reveals is precisely the early and still fluid ideological form of Mahāmudrā during its transmission from India to Tibet.

In accordance with the tenets of the bKa’-brgyud lineage of Tibetan Buddhism, the Mahāmudrā practice can be subdivided into two distinct categories: Sūtra Mahāmudrā and Tantra Mahāmudrā (bSod rnams tshe ring 1998, pp. 399–403; Khro ru tshe rnam 2002, p. 262; Brunnhölzl 2014, p. 151). This dual classification facilitates comprehension of the Mahāmudrā practice. The actual founder of the Dwags po bKa’ brgyud lineage, Gampopa (sGam po pa bSod nams rin chen, 1079–1153), formalized this dual classification.1

In the 15th century, the famous Tibetan historian and scholar ’Gos Lo tsā ba gZhon nu dpal (1392–1481) wrote a chapter titled “The Mahāmudrā (Phyag rgya chen po’i skabs)” in the Blue Annals (Deb ther sngon po), in which he stated about the Mahāmudrā doctrine, “which seals all the meditative and religious practices, from the Prātimokṣa, which is the foundation of the Doctrine of Buddha, to the Guhyasamāja Tantra (“GST” hereafter. Deb sngon, p. 1173; Roerich 2016, p. 839). While this statement encompasses a broad spectrum of Buddhist practices, the specific mention of the GST, a foundational scripture of Indian Buddhist tantra, suggests that such tantric sources were not regarded as external to the Mahāmudrā tradition, but rather as integral to its doctrinal range as understood by ’Gos Lo tsā ba.

In the book History of Tibetan Esoteric Buddhism, bSod rnam tshe ring (1962–present) makes the following statement: Mahāmudrā is characterized by a wealth of content, comprehensive theoretical principles and a distinctive practice. They integrate the concept of emptiness held by the Mādhyamika with the practice methods in the anuttarayoga class of tantras, including the GST, Hevajratantra, Cakrasaṃvaratantra, Mahāmudrātilaka, and Kālachakratantra (bSod rnams tshe ring 1998, p. 403). This assertion indicates that the Tibetan Mahāmudrā practice possesses the characteristic of integrating both sūtras and tantras.

In order to gain a deeper understanding of the original attributes of Mahāmudrā practice prior to its reinterpretation by Gampopa, it is essential to commence with the instance of the transmission of Mahāmudrā practice from India to Tibet, thereby comprehending its development process. The Indian Buddhist scholar Jñānakīrti (Ye shes grags pa, around the late 10th century to the early 11th century) composed the TA (Lu 2018, 2019; Zhang 2019). This is a treatise that integrates related sūtras, tantras, and commentaries, rather than being a special commentary on a particular sūtra or tantra. In terms of textual structure, the TA organizes the stages of practice according to different faculties, and its content is about the theory and oral instructions (man ngag) of Mahāmudrā practice.

The purpose of analyzing the textual composition is to understand the teachings that served as an antecedent for the Tibetan tradition. Previous scholars were concerned with the chapter titled “Irreversible Sign in Prajñāpāramitā” (Shes rab pha rol tu phyin pa’i phyir ma ldog pa’i rtags) in the TA, which is related to Sūtra Mahāmudrā (Mathes 2006, pp. 201–27; Mathes 2008, pp. 34–45, 367–410). The chapter “Irreversible Sign in Prajñāpāramitā” comes after the chapter “Practices of Secret Mantra Approach” (gSang sngags kyi sgo’i spyod pa) in the TA and is shorter in length compared to the latter. The “Practices of Secret Mantra Approach” occupies a large portion of the TA, and it is seldom discussed (Brunnhölzl 2014, pp. 184–88). From its title, we know that the content of this chapter is about mantra (gsang sngags) practice, or it can be considered as the Tantra Mahāmudrā teaching. Due to the complexity of the transmission lineages, the Tantra Mahāmudrā teaching has always been a subject of controversy. Normally, the teaching is about attaining wisdom through the practice of [vital] energy (prāṇas), channels (nāḍīs), and essences (bindus), and experiencing great bliss and luminosity (bSod rnams tshe ring 1998, pp. 402–3). Is this consistent with the teachings in the TA? Examining the practice methods in the “Practices of Secret Mantra Approach” can reveal the early context of the Mahāmudrā teachings that were directly transmitted from India to Tibet.

2. Summary of the “Practices of Secret Mantra Approach”

The TA, full name De kho na nyid la ’jug pa zhes bya ba bde bar gshegs pa’i bka’ ma lus pa mdor bsdus te bshad pa’i rab tu byed pa (*Tattvāvatārākhyāsakalasugatavācasaṃkṣiptavyākhyāprakaraṇa, Realizing Reality: Brief Explanation of All Sugata’s Words), currently exists only in Tibetan translation included in the collection rGyud or rGyud ’grel of the Tibetan Tripiṭaka-Tanjurs. The translators are Padmākaravarman and dGe slong Rin chen bzang po (958–1055).

There are six editions in Tibetan:2

(1)

B: dPal spungs edition of the Phyag chen rgya gzhung, vol. hūṃ, fols: 320v4–377r3.

(2)

C: Co ne bsTan ’gyur, No. 3709, rGyud, vol. Tsu, fols. 38v6–75v6.

(3)

D: sDe dge bsTan ’gyur, No. 3709, rGyud, vol. Tsu, fols. 39r2–76r4.

(4)

G: gSer bris ma bsTan ’gyur, No. 2539, rGyud ’grel, vol. Nu, fols. 52r1–99r6.

(5)

N: sNar thang bsTan ’gyur, No. 4532, rGyud ’grel, vol. Nu, fols. 42v7–84v1.

(6)

P: Peking bsTan ’gyur, No. 4532, rGyud ’grel, vol. Nu, fols. 42v2–84v2.

The text of the TA comprises about 37 folios and is divided into two parts. The first part is the general doctrines (D39r2–43r6, about 4 folios) divided into three chapters, namely, the “Detailed Definition of Classifications,” (dBye ba’i rnams par nges pa)3 the “Detailed Definition of Ācārya’s Methods” (sLob dpon gyi tshul zhes bya ba’i rnam par nges pa) and the “Common Teachings” (Thun mong bstan pa zhes bya ba). The second part is specific doctrines (D43r6–76r4, about 33 folios), also divided into three chapters: the “Practices of Secret Mantra Approach,” (about 16 folios) the “Irreversible Sign in Prajñāpāramitā” (about 11 folios) and the “Path Free from Attachment” (’Dod chags dang bral ba’i lam, about 6 folios). In the text, the three chapters of the first part and the second part are all called “Chapter One,” “Chapter Two,” and “Chapter Three”.

According to Chapter One in the first part of the TA, the outline of the “Practices of Secret Mantra Approach” is as follows:

In this regard, those with superior faculties among those who are engaged in practicing the approach of secret mantra [have such superior faculties,] because they have great compassion and little disturbances, and they are wholly dedicated to cultivating the yoga of being one with mahāmudrā.

As for those with intermediate faculties, even though they definitely have no conviction to emptiness, they are overpowered by disturbances. Therefore, they are wholly dedicated to cultivating the yoga of being one with a jñānamudrā, which has the nature of emptiness.

Although those with inferior faculties who are also dedicated to emptiness, their continuum of mind is overpowered by the fierce disturbances of attachment and so forth. Because of that, they rely on (are endowed with) the yoga of samayamudrā and karmamudrā according to the circumstances.4

This foretells that the content of the chapter is the instruction of the yoga (union) of being one with (inseparable from) Mahāmudrā for the practitioners (yogins) with superior faculties. Here, the “superior faculties” are divided into three kinds, respectively, as those of inferior, intermediate, and superior. In other words, the three kinds of faculties in this chapter are actually the top three of the “nine faculties in three yānas” in the TA, i.e., the superior, intermediate, and inferior sub-classes of superior faculties.

The section discussing the superior sub-class of superior faculties is long (D43r6–55v6, about 12 folios), while the intermediate (D55v6–57v7, about 2 folios) and inferior sections (D57v7–59r5, about 2 folios) are short. Therefore, the superior section is core, and the content of the intermediate section and inferior section will not be discussed for the moment.

Regarding the teachings on the superior sub-class of superior faculties, its explanation comprises a series of questions and answers. The initial question to face is “What is Mahāmudrā non-dual yoga?”, and the answer is as follows:

That will be explained:

“The union of skillful means and insight is nothing but cultivation for the supreme yogins,

The Buddhas call it the meditation of Mahāmudrā yoga.”

Cultivating the self-nature of skillful means and insight itself is cultivation on Mahāmudrā non-dual yoga.5

This is not a direct explanation of what the meaning of Mahāmudrā non-dual yoga is, but it is an answer to the method of practice. This signifies that the practice of the self-nature of skillful means (thabs, *upāya) and insight (shes rab, *prajñā) is tantamount to the practice of Mahāmudrā. It is evident that skillful means and insight occupy a pivotal role within the Mahāmudrā non-dual yoga system.

A further question arises: “What is insight? And what is skillful means? How are they cultivated? Together? Or separately?” (D43v1–2) These are all things that need to be paid attention to during the practice of the superior faculties. To address these issues, Jñānakīrti divides his answer into two parts: first, an explanation of insight; second, a discussion of skillful means.

From the above contents, we can summarize the basic framework of the “Practices of Secret Mantra Approach”: according to the capacities, the practitioners are divided into three kinds. Attach importance to the practice of superior faculties, and the most important thing is the instruction of skillful means for the superior of superior faculties. The Mahāmudrā teaching advocates the practice of insight and skillful means at the same time, which represent the practice theory and practice methods, respectively. The theory of practice mainly discusses the thoughts of emptiness. Most of the practice methods can correspond to the Tattvasaṃgrahatantra and the GST, including maṇḍala contemplation.

3. Insight (Prajñā): Absence of Fabrication Means Emptiness

There are some brief explanations for the insight section (D43v1–45v3). The connotation of insight is expressed as “the realization of the absence of fabrication in all phenomena.” (D43v2–3) This further asks what the “absence of fabrication” is, and the answer is “emptiness (stong pa nyid)”. The subsequent questions are discussed around emptiness, involving some views between Madhyamaka and Yogācāra, thus determining that “fully comprehending absence of fabrication means abiding in the insight”. (D43v3–45v3).

In the discourse from the section on the practice theory of insight, many issues are raised and discussed. For instance, it is necessary to understand that “All words operate for the sake of removing all kinds of superimposition,” (D43v4) and to distinguish it from the emptiness spoken by the non-Buddhist, Yogācāra, etc. After raising an issue, Jñānakīrti would first summarize it with a gāthā (verse, usually of four lines), then explain the gāthā as a detailed answer, and so on. Some of these gāthās come from their own sources, and some are unknown.

In this section, the core ideas should come from Nāgārjuna. Nāgārjuna (ca. 2nd–3rd century) is one of the most influential philosophers in the history of Indian Buddhism and is regarded as the founder of the Madhyamaka school. In explaining the relationship between emptiness and absence of fabrication, and the relationship between removing all superimposition and reality, Jñānakīrti quoted some gāthās from Nāgārjuna’s Mūlamadhyamakakārikā (abbreviated as MMK). The MMK is the core canonical text of the Mādhyamaka school. The translation and annotation of the MMK can be categorized into four main trajectories: Sanskrit original text–Chinese translation–Tibetan translation–modern retranslation. Spanning more than 1600 years, this process has given rise to an extensive corpus of texts that transcend traditions and language families. Here, only the early translations are taken as examples. In 409, Kumarājīva 鳩摩羅什 (343–413 or 350–409), one of the three most renowned translators in Chinese history, translated the four-volume Zhonglun 中論, which stands as the earliest complete version. It is based on the commentary by Indian Buddhist master Piṅgala (ca. 320–400). Kumarājīva translated the MMK appending Piṅgala’s extended prose commentary. In the early ninth century, Jñānagarbha (distinguished from the homonymous Madhyamaka master in the eighth century, (Ye 2011a, p. 18)) and Cog ro Klu’i rgyal mtshan (ca. 8th century) rendered the Mūlamadhyamakavṛttyakutobhayā (D3829) into Tibetan, which contains the complete MMK. This version of MMK was later revised and compiled into a stand-alone edition (D3824) within the bsTan ’gyur. Since the modern era, through Sanskrit manuscripts and Tibetan documents from Nepal and Tibet, there have emerged numerous critical editions, translations, and studies in various modern languages, including Chinese, English, German, Japanese, etc., in order to reconstruct the original appearance of the MMK (Ye 2011a, pp. 18–21). In 2011, Ye Shaoyong published a new critical edition of the MMK, in which he used the Tibetan translation by Cog ro Klu’i rgyal mtshan (Ye 2011b). Two years later, an English translation of the MMK was published (Siderits and Katsura 2013). The ancient and modern Chinese translations of these two verses are largely consistent. The primary differences between them are formal: The ancient Chinese translation by Kumārajīva takes the form of a five-character regulated quatrain, whereas the modern Chinese translation by Ye Shaoyong is a seven-character regulated quatrain. The latter is more readily understandable to modern readers. And, based on faithfulness to the Sanskrit original, the modern English translation by Siderits and Katsura is rendered concisely and smoothly. All of aforementioned versions have been widely utilized in academic circles.

Here, Jñānakīrti quotes Nāgārjuna’s MMK. One example is about the 6th gāthā in Chapter 15 “An Analysis of Intrinsic Nature” of the MMK (15.6 hereafter) (

Table 1. MMK 15.6.

Language	TA	MMK 1
Sanskrit	(none)	svabhāvaṃparabhāvaṃca bhāvaṃcābhāvaṃeva ca | ye paśyanti na paśyanti te tattvaṃbuddhaśāsane ||
Tibetan	(D44r2) rang gi dngos dang gzhan dngos dang | | dnogs dang dngos med nyid dag ni | (| de dag sangs rgyas bstan pa la |2) | gang zhig mthong dang mi mthong de | | sangs rgyas bstan la de nyid min (| | yang dag mthong ba ma yin no |3)	gang dag dngos nyid gzhan dngos dang || dngos dang dngos med nyid lta ba || de dag sangs rgyas bstan pa la || de nyid mthongs ba ma yin no ||
Chinese	(new translation) 自性與他性，實有與無有， 彼等於佛說，無論見非見， 並非真佛說，不是清淨見。	(Kumārajīva’s translation of the Zhonglun) 若人見有無，見自性他性； 如是則不見，佛法真實義。 4
		(Ye’s translation) 某些人等若觀見，自性、他性及有、無， 彼等人則不觀見，佛陀教法中真實。
English	(new translation) Self-entity and other-entity, entity and non-entity, these are not taught by Buddha. What is perceiving and what is non-perceiving, [these are] not the reality of the Buddha’s teachings, nor correct perceiving.	Intrinsic nature and extrinsic nature, existent and nonexistent— who see these do not see the truth of the Buddha’s teachings. 5

¹ The Sanskrit, Tibetan translation and Chinese translation, see (Ye 2011b, pp. 238–39). ² GNP om. ³ GNP om. ⁴ CBETA 2024.R2, T30, no. 1564, p. 20a23–24. ⁵ (Siderits and Katsura 2013, p. 159).

):

From the perspective of the Tibetan translation, the gāthā 15.6 can be considered as having two versions, a four-sentence version and a six-sentence version. The G, N, and P editions of the gāthā are the regular four-sentence version. The other three editions fully contain these four sentences of 15.6 in the MMK, with slight changes in wording, but the specification has become a six-sentence version. The editors of the three editions such as the P may have considered the regularity of the gāthā specifications, or they may have had other considerations, which differ from the other editions. In the following text, the Degé and Peking editions are used as representatives of the six-sentence version and the four-sentence version, respectively.

By comparing the Sanskrit and Tibetan editions, some situations can be discovered. Let us first examine the sentence structure. The first two lines of the Sanskrit MMK 15.6 employ four concepts, namely svabhāva (self-nature, self-entity, 自性), parabhāva (other-nature, other-entity, 他性), bhāva (being, entity, 實有, 有), and abhāva (non-being, non-entity, 無), which are connected in a parallel construction. In the Tibetan translation of the MMK, these two padas (lines) contain the element “gang… lta ba” (who see). This is actually part of the third pada that has been translated in advance. In the TA, the structure of these two padas corresponds more closely to that of the Sanskrit MMK. The only superfluous elements are function words such as “dag ni”. Next, consider the specific vocabulary used in the Tibetan translations. The Tibetan translation equivalent of “svabhāva”, namely “dngos nyid”, corresponds to “rang gi dngos” (entity of self) in the TA. This is an interpretive substitution and has been translated into a three-syllable phrase, which might be aimed at meeting the requirements of Tibetan translation prosody. In the TA, the other three key terms maintained the same Tibetan translation: gzhan dngos (parabhāva), dngos (bhāva), and dngos med (abhāva).

More significant differences are found in the last two padas. The third pada of this verse in the Degé edition of the TA is identical to the third pada in the Tibetan translation of the MMK. But in the Peking edition of the TA, there is no direct equivalent of this pada. However, it can be associated with the last pada of this edition, namely “sangs rgyas bstan la de nyid min”. And this pada also exists in the Degé edition of the TA, being the fifth pada. The Sanskrit counterpart of this pada is the fourth one. In comparison with the Tibetan translation of the MMK, the fourth pada (gang zhig mthong dang mi mthong de) in the Degé edition of the TA preserves the parallel construction of “see and not see” (ye paśyanti na paśyanti) in the Sanskrit original. This form, which corresponds to the syntactic structure of the Sanskrit sentence, demonstrates the translators’ fidelity to the Sanskrit text. In the fifth and sixth padas, Jñānakīrti divided negation into two steps. First, “sangs rgyas bstan la de nyid min” negates the consistency with what the Buddha taught. Then, “yang dag mthong ba ma yin no” negates the correctness of the previous view. This is encapsulated in the phrase “tattvaṃ buddhaśāsane” in the original Sanskrit text. Jñānakīrti’s rephrasing represents an emphasis-based dissection. This is to say, the last two padas of MMK 15.6 were elaborated upon in the quotation within the Degé edition of the TA.

This kind of “quotation” by Jñānakīrti in the definition of modern scholars may belong to “repeat,”, that is, the “text reuse” without a clear indication of the source (Hugon 2015, pp. 454–55). It represents a form of expression that occupies an intermediate position between quotation and commentary. The modifications made by Jñānakīrti when citing the MMK, such as reordering of word sequences, insertions of explanatory expansions, and the dissection of conclusions, reveal his recreative and adaptive utilization of the original text. This approach follows the theory of huayong 化用 (creative reuse), a term derived from Chinese rhetoric studies. This quotation not only preserves the crucial structure but also makes adjustments in terms of diction and logic to suit the argumentative objectives of the TA. This form of creative reuse not only reflects Jñānakīrti’s comprehension and reinterpretation of the ideas within the MMK but also mirrors the “dynamic” tradition of employing Madhyamaka scriptures by different Buddhist scholars.

Another similar example is the 5th gāthā in Chapter 25 “An Analysis of Nirvāṇa” of the MMK (

Table 2. MMK 25.5.

Language	TA	MMK 1
Sanskrit	(none)	bhāvaś ca yadi nirvāṇam nirvāṇaṃsaṃskṛtaṃbhavet | nāsaṃskṛto vidyate hi bhāvaḥkvacana kaścana ||
Tibetan	(D44v7) gal te mya ngan ’das dngos yin | | mya ngan ’das pa ’dus byas ’gyur | | dngos po gang dang gang na yang | | ’dus ma byas pa yod ma yin |	gal te mya ngan ’das dngos na || mya ngan ’das pa ’dus byas ’gyur || dngos po ’dus byas ma yin pa || ’ga’ yang ji ltar yod ma yin ||
Chinese	(new translation) 若涅槃實有，涅槃成有為。 無論何處有，皆非無為法。	(Kumārajīva’s translation of the Zhonglun) 若涅槃是有，涅槃即有為； 終無有一法，而是無為者。 2
		(Ye’s translation) 如果涅槃是事物，涅槃則成有為法。 無為之物無有故，無論何者於何處。
English	(new translation) If nirvāṇa were an entity, nirvāṇa would be conditioned. Wherever any entity [exists], the unconditioned does not exist.	And if nirvāṇa were an existent, nirvāṇa would be conditioned, For never is there found and any existent that is not conditioned. 3

¹ (Ye 2011b, pp. 452–53). ² CBETA 2024.R2, T30, no. 1564, p. 35a14–15. ³ (Siderits and Katsura 2013, p. 293).

):

This gāthā emphasizes that all entities are conditioned in the world; however, nirvāṇa is unconditioned. From the Tibetan translations of the gāthā, the first two padas are slightly different. The main difference lies in the wording of the last pada. In the MMK, the phrase “’ga’ yang ji ltar yod ma yin” is used, emphasizing that “nothing can exist in such a manner”; whereas Jñānakīrti modified it to “’dus ma byas pa yod ma yin”, directly negating the existence of “a dharma that is not born from causes and conditions”. This alteration shifts the semantic focus from the original verse’s broader negation of all self-nature existence to a more specific denial of the possibility of “non-conditionally originated dharma”, concentrating on the non-existence of “non-conditioned dharma”.

The above two instances show that Jñānakīrti was significantly influenced by Nāgārjuna’s MMK. He supported Nāgārjuna’s perspective and employed it as a foundation for his own line of reasoning. In the TA, Jñānakīrti dispelled practitioners’ attachment to any self-natured entity or ultimate dharma by negating the existence of “substantially existent entities” and “unconditioned dharmas”. In this way, the Mahāmudrā practice can revert to the state of primordial awareness, where all dharmas are empty, free from fabrication, and do not abide in any characteristics. Additional similar examples exist. In the “Irreversible Sign in Prajñāpāramitā” of the TA, Jñānakīrti also huayonged (creatively reused) the first and second gāthās in Chapter One “An Analysis of Conditions” of the MMK in order to elaborate on the concept of transcending the limitations of mind-only thinking (Lu 2019, pp. 299–307). Jñānakīrti’s quotation of the MMK in the TA not only inherits the original text’s ideas but also serves as an indication of the recreation and adaptive utilization of classical content during the dissemination of the Madhyamaka school in the Tibetan context. In essence, he can be defined as a paṇḍita who held a Madhyamaka philosophical position. And this kind of “huayong” (or “text reuse”) was relatively common in the late Buddhist period in India.6

4. Skillful Means (Upāya): Focus on the Maṇḍala and the GST

The skillful means section is longer (D45v3–55v6). Here, the extension of skillful means is defined as the three types of Buddha-mother’s great compassion (snying rje chen po) distinguished by their different meditative focuses. The first is that which arises in relation to sentient beings. The second is that which arises in relation to phenomena. And the third is that which arises from meditation on the non-entity of all phenomena and beyond. Subsequently, it is stated that skillful means and insight are the essence of compassion. The yogins become the nature of great compassion by having become the nature of all entities through the stages of meditation. Thereafter, they become the nature of Mahāmudrā. Then, the instructions for practicing skillful means and insight are followed. The specific instructions are to explain the complex structure of the maṇḍala visualization, combined with a number of tantras and sūtras to elaborate, especially the GST, and ending with a long gāthā.

Jñānakīrti focused on proclaiming the practice instructions (D46r2–50r6). The yogins are instructed to generate the Deity Yoga (lha’i rnal ’byor) by visualizing the various deities one by one, thus forming a maṇḍala. Deity Yoga is the essence of Vajrayāna (Esoteric Buddhism) practice. It is a profound approach designed to transform the yogin’s ordinary body, speech, and mind directly into the enlightened body, speech, and mind of a buddha or bodhisattva (i.e., the meditational deity) through visualized self-identification with its sanctified virtues. In his magnum opus The Great Treatise on the Stages of Mantra (sNgags rim chen mo) (CBETA 2025.R1, B10, no. 68), the well-known Tibetan Buddhist philosopher Tsongkhapa (1357–1419) established Deity Yoga as the defining feature distinguishing Vajrayāna from Sūtrayāna, systematically expounding its philosophical foundations, practical methods, and sequential stages (Power 2005).

Jñānakīrti expounded on the meditation practice of the Deity Yoga with a specific emphasis on the Generation Stage (utpattikrama). According to the teachings, yogins are required to visualize themselves as the deity immediately. Afterwards, they need to realize that both themselves and all phenomena inherently possess the characteristics of skillful means and insight. This involves a transformation of the Five Aggregates (phung po lnga) and the Five Fundamental Afflictions (nyon mongs pa lnga), which are transformed into the Five Buddha Families and their corresponding wisdoms. The text clearly links the Five Aggregates to the Five Dhyāni Buddhas: The form aggregate (gzugs kyi phung po) corresponds to Vairocana Buddha (rNam par snang mdzad, for the transformation of ignorance), the feeling aggregate (tshor ba) corresponds to Ratnasaṃbhava Buddha (Rin chen ’byung, for the transformation of arrogance), the perception aggregate (’du shes) corresponds to Amitābha Buddha (’Od dpag med, for the transformation of desire), the mental formations aggregate (’du byed) corresponds to Amoghasiddhi Buddha (Don yod par grub pa, for the transformation of envy), and the consciousness aggregate (rnam par shes pa) corresponds to Akṣobhya Buddha (Mi bskyod pa, for the transformation of hatred). The process of transforming is a key aspect in tantric practice, revealing the fact that the Buddha nature is inherently present in all sentient beings and phenomena.

A maṇḍala encompasses elements such as the pure realm of the deity, consorts, and attendant retinues, which is a sacred and symbolic configuration. Through the composition of the deities in the maṇḍala, it is evident that the Vajradhātu Maṇḍala is introduced here by Jñānakīrti. The deities include the Five Dhyāni Buddhas (as mentioned above), the Four Gates of Liberation, the Sixteen Bodhisattvas, the Four Pāramitās and their consorts, and the Four Gatekeepers, totaling 37 deities of the Vajradhātu Maṇḍala (Ruan 2012; Dangcuo 2019). These are fully presented in this section, and the names of the deities differ only slightly from those found in the Tattvasaṃgrahatantra translated into Chinese, reflecting a different transmission. In the explanation of the maṇḍala, there is a long gāthā containing a series of scripture titles, such as the De nyid bsdus pa (*Tattvasaṃgrahatantra), Dam tshig chen po’i rgyud (*Samayatantra), dPal ldan rdo rje rtse mo (*Vajraśekharatantra), dPal mchog dang po’i rgyud (*Paramādikalpatantra), dPal ldan ’dus pa’i rgyud (GST), etc., which is supposed to express admiration. This also reflects the prevailing text of that time.

The maṇḍala is followed by extensive quotations of the GST. For example, in proclaiming the nature of Bodhicitta (byang chub kyi sems), Jñānakīrti quoted the following:

The Blessed One, the Master who has directly realized the Secret Assembly of the vajra body, speech, and mind of all Tathāgatas, how should he be viewed by all Tathāgatas? Son of noble family, he should be viewed as the mind of enlightenment (byang chub kyi sems) by all Tathāgatas and all Bodhisattvas.

Why is this? The master is equal to the mind of enlightenment and is indivisible into two aspects.

Son of noble family, in brief, all the Buddhas and Bodhisattvas who abide, live, and sustain [beings] in the world systems of the ten directions have come in the three times. Having made offerings to that master with the offerings of all the Tathāgatas, they also go to that buddha field and utter these vajra syllables: “We are the fathers of all the Tathāgatas. We are the mothers of all the Tathāgatas.” Then, “We are the teachers of all the Tathāgatas.”

It is like this. Son of noble family, all the blessed Buddhas who reside in the ten directions, the merit of the vajra body, speech, and mind of all those blessed Buddhas is surpassed by the merit of a single pore of the master.

Why is this? Because, son of noble family, the mind of awakening is the essence of the wisdom of Buddhahood. It is the source of all the qualities of the Buddha, from the state of being the place of birth up to the source of the wisdom of omniscience.7

The above quotation is from Chapter 17 of the GST. The GST is considered one of the central scriptures of the Anuttarayoga Tantra in Tibetan Buddhism. Classified within the “Father Tantra” category of this highest yoga tradition, it emphasizes skillful means as its primary path. Actually, it occupies a lofty status as the “King of Tantras” within the Esoteric Buddhist system. The GST is available in Sanskrit manuscripts, Tibetan translations, and Chinese translation. Typically, scholars hold that the extant Tibetan translations of the GST are slightly more reliable than the Sanskrit versions. The Tibetan translations date back 500 to 1000 years earlier than the various Sanskrit manuscripts (Hackett 2017, pp. 168–69).

The textual expression found in the Tibetan translation of the GST closely corresponds to this quoted content.8 But in his TA, Jñānakīrti adapted the quotations from the GST through multi-layered modifications. Between the two texts, there exist not only formal differences such as lexical substitutions, additions, omissions, and sentence order adjustments, but also several semantic/doctrinal fine tunings. The most striking divergence lies in the transformation of key terminology: The tantric-core expression “Master who has been empowered” (dbang bskur ba’i slob dpon) in the GST is replaced by “Master who has directly realized” (rtogs pa’i slob dpon) in the quotation. This revision downplays the ritualistic dimension unique to tantra while accentuating the path of prajñā realization—aligning with Jñānakīrti’s goal of bridging exoteric and esoteric teachings. When defining “viewing the master as Bodhicitta (mind of enlightenment),” the GST instructs disciples to perceive the guru as “vajra-bodhicitta” (rdo rje nyid du blta), emphasizing its tantric essence. The quotation strips away the “vajra” symbolism, simplifying it to the generic “view as Bodhicitta” (sems su blta). This demystification renders the doctrine more accessible to exoteric scholars. Regarding the relationship between Master and Bodhicitta, the GST binds them through the vajra symbol: “This vajra-bodhicitta and this master are non-dual and undifferentiated.” The quotation removes this tantric marker, stating only “The master is equal to the mind of enlightenment and is indivisible into two aspects”—a conceptual streamlining that facilitates exoteric-esoteric synthesis. Another notable adjustment concerns the concept of “buddha field”: The original’s “go to one’s own buddha field” (rang gi sangs rgyas kyi zhing du ’gro), implying the tantric goal of innate accomplishment in this very life, is generalized in the quotation to “go to that buddha-field” (sangs rgyal kyi zhing der ’gro). Thus, from the perspective of the author, it can be seen that Jñānakīrti transformed the tantric logic of the GST into a framework of “bodhicitta is the wisdom of emptiness” that is understandable within the context of exoteric Buddhism.

These differences may also reflect some of the following alternative situations: (1) the genealogical distinctions among the base readings, (2) the abridged translation or paraphrased rewriting when citing in treatises, and (3) translation batches and revision levels. The GST was first rendered into Tibetan by Rin-chen-bzang-po in collaboration with the Indian paṇḍita Śraddhākaravarman; it was re-edited by the translator Chos rje dpal (1197–1264), producing the version that is enshrined in today’s Degé Kangyur. The Tibetan translation of the TA was completed somewhat later (in the mid- to late 11th century) under the direction of Rin-chen-bzang-po together with another Indian paṇḍita. By then, the editorial environment and the Sanskrit exemplars differed from those used for the initial translation of the GST. The Sanskrit manuscripts of the GST in circulation in India and Tibet at the time were not of a single version. Therefore, these textual variations came into being.

It can also be referred to vol. 6 in the Chinese translation Miji benxu by Dānapāla of the Northern Song Dynasty (CBETA 2025.R1, T18, no. 885, p. 503b1–18), which is completed by ca. 1002. The Chinese translation is not a literal word-for-word rendering but contains a substantial number of explanatory passages (Matsunaga 1978, p. VI). When used to verify version differences, its reference value is limited. Therefore, this shall not be discussed further here.

Following the above quotation, there are two more long quotes from the GST that talk about Bodhicitta. It can be seen that GST occupies an important place in this section, and the concept Bodhicitta is the keyword here.

After this, the scriptures that are explicitly quoted are the Tattvasaṃgrahatantra, the sDom pa nyi shu pa (*Bodhisattvasaṃvaraviṃśaka, by Candragomin), the Chos mngon pa’i mdzod (*Abhidharmakośa), and the De bzhin gshegs pa’i gsang ba’i mdzod kyi mdo (*Tathāgataguhyasūtra).

In the intermediate section and inferior section of superior faculties, only the quotations explicitly mentioned in the Ye shes phyag rgya mdo, and the rest of the content is mostly gāthās, whether there is huayong remains to be examined.

Considering the relationship between sūtra and śāstra, to some extent, the main content of the superior sub-class of superior faculties in the “Practices of Secret Mantra Approach” of the TA is a short commentary on the extended version of the Vajraśekharatantra. Because much of the content of this section is derived from the Vajradhātu Maṇḍala of the Tattvasaṃgrahatantra and the GST. As some academic traditions hold that the Tattvasaṃgrahatantra is Chapter 1 of the extended version of the Vajraśekharatantra, and the GST is Chapter 15 (Ruan 2012; Dangcuo 2019; Li 2022). Actually, this composition shows that the Mahmāudrā teaching of Jñānakīrti was established in the period of popularity for the tantras such as the Tattvasaṃgrahatantra and GST.

5. Conclusions

Understanding the thoughts of predecessors from the position of later generations would create an incompatible situation. In the understanding of Indian Mahāmudrā, we shouldleave behind the mature theory of the bKa’-brgyud lineage after the 12th century, return to the Indian tradition, and re-examine the source of classical thought.

In the late period of Indian Buddhism, the Mahāmudrā teaching reached the stage of Jñānakīrti (around the 11th century) and showed the guiding ideology of practice based on the “nine faculties in three yānas” framework from the top down through the TA. The core chapter of the TA is the “Practices of Secret Mantra Approach”, but this part has already integrated the thought of the Madhyamaka and tantric rituals. This part may have been the precursor to what later became the Tantra Mahāmudrā. However, at that time, there were no specific expressions to distinguish the types of Mahāmudrā such as Sūtra Mahāmudrā or Tantra Mahāmudrā.

The “Practices of Secret Mantra Approach” of the TA focuses on the teachings of the superior sub-class of superior faculties, emphasizing the practice of insight and skillful means together, i.e., combining the thought of emptiness with the maṇḍala visualization. It is worth noting that the relevant content from the extended version of the Vajraśekharatantra appears in this chapter. The Vajraśekharatantra is not particularly emphasized in the bKa’-brgyud lineage. This means that, in addition to paying attention to the close relationship between the GST of the anuttarayoga class in Tibet and the Mahāmudrā teaching, attention should also be paid to the Tattvasaṃgrahatantra of yoga class. The analysis of the text composition reflects the early form of the Mahāmudrā teachings fromwhen they were introduced into Tibetan Buddhism, that is, the development process from yogic teachings to supreme yogic teachings.