Research lines opened by Natarajan's practitioner text and the site's non-tattva sections, carried forward rather than resolved. Each is stated as a question with the next concrete step named, not as a direction.
The claim, from phase 10: a tradition's safety doctrine fails not because it is missing but because it is filed away from the practice it governs, and the distance is measurable in pages. Two corpora, fifty years apart, no contact, same failure. [n=2]
Why it might be general. Safety material is written by whoever is thinking about safety, and gets filed with the rest of their writing. Nothing in an anthology, a prompt-response corpus, or a manual series makes a constraint stated in one document binding in another.
Next step, concrete. ⭐ Test it on a corpus this vault already holds and that has nothing to do with Hanumān. The Bodri ingest is the obvious candidate — a practitioner text with extensive cautions. Measure the page-distance between each caution and the practice it governs. If the pattern holds at n=3 across unrelated traditions, it stops being a finding about this build and becomes a claim about how practice literature is assembled.
⚠ What would kill it: finding a tradition that routinely attaches its gates at the point of use. The five manuals the Carit names are the first place to look — if they carry adhikāra material inline, the Aṅka's ritual essays stripped it in transmission, which is a worse and more interesting finding.
The claim: every progression in both corpora ends in a described condition, assessed by the person hoping to have reached it. The Carit ends its subject's arc in a job — removing the obstacles in their practice, day and night — and a job can be failed.
Why this is the phase's most usable finding. A state-terminal progression has no stopping criterion by construction: there is nothing after arrival to check yourself against. A role-terminal one does — you are doing the work or you are not.
Next step, concrete. ⭐ Rewrite one of the corpus's existing progressions with a role at the summit instead of a state, and see what breaks. The twelve-week curriculum is the right size to try it on. The prediction: the assessment apparatus becomes unnecessary rather than improved, because a role supplies its own evidence.
🧭 The open question underneath: is the role's content transferable, or does it depend on there being an unbounded population of strangers with obstacles? A practitioner with no such population has a summit they cannot occupy.
The claim, at n=3: this corpus's usable formats survive only inside worked examples, never in the essays that theorise the same subject. The report format sits in a narrative chapter while four envoy doctrines name no deliverable. The scarcity speech sits in a bridge-building account while the influence material imports Cialdini from outside. The observer-effect argument sits in a council scene and is stated as a general principle nowhere.
Why it matters more than the individual findings. It predicts where to look. A corpus's stories contain its artefacts; its essays contain its principles; and readers who sample essays acquire doctrine and no formats.
Next step, concrete. ⭐ Run it as a deliberate extraction pass rather than as a by-product. Take the Carit's thirteen chapters and read them only for reproducible formats — orderings, checklists, sequences, thresholds — ignoring everything doctrinal. The three found so far were all incidental discoveries; a pass designed for it should return more.
⚑ And the same pass should be run on the extension corpus's own narrative material, which this build has largely read for its claims.
Carried forward from phases 7 and 9 and sharpened by phase 10.
Phase 7 asked how far the technology separates from its object — one document makes the deity a variable while the tradition's strongest devotional material makes exclusivity the test. Phase 9 found six protocols built on self-installation ("you're not asking for help, you're becoming the solution") and none on petition, for a figure whose most widespread form is defined by being asked.
Phase 10 supplies the evidence neither had. ⭐ The sahasranāma's operational block is entirely petitionary and entirely specific: pregnancy, debate, anger, sorcery, sick children. That is what a thousand years of households actually brought to this figure, and it sits inside a text whose frame is total cosmological identification.
Next step, concrete. ⭐ Set the operational block beside the corpus's protocol list, item for item, and ask which needs the object to be a person. Remover of afflictions in pregnancy needs somebody to ask. Timeline template does not. The line between them is the answer to phase 7's question, and it is now drawable from evidence rather than from argument.
The narrowest thread and possibly the most portable.
Every other instance of Rāma valuing Hanumān in this corpus is retrospective — the debt, the embrace, the public tribute, all after performance. The sahasranāma is the exception: a thousand names spoken before the mission is given, to a servant who has done nothing yet, and the text claims the naming is what produced his recognition of who Rāma was.
The competing explanation is genre. A sahasranāma has to be spoken by someone to someone at some point, and the first meeting is the natural slot. The build cannot rule it out.
Next step, concrete. ⭐ Find a second instance. Any sahasranāma or comparable naming-text placed before a commission rather than after an achievement. One more case makes it a structure; none makes it a compositional convenience — and the distinction decides whether assess first, assign second is a technique worth building on or a nice reading of a set-piece.