Under author review · MIT · 31 tests · 6,415 assertions
Retrieval Context Provenance Auditor
Which retrieved information influenced this decision, and how trustworthy was its supply chain?
Previous pull request #164 ↗SHPBL agent evidence series
SHPBL discovered and composed these products from capability intent across its library and CMPSBL, a sister project by the same developer. Each one is a separate MIT package with its conformance tests in the open. The last six were composed on top of two practitioners’ public security tooling, credited below.
How these twelve came to exist
Set one · SHPBL × its own library
The first six were composed by combining SHPBL with its own capability library and CMPSBL — governed capability intent on both sides of the join. Nothing external went in; the software that came out did not exist before the combination. That is the whole method: capability composition producing an artifact neither input contained.
Set two · SHPBL × two Exchange authors
The second six ran the same method with a different second input: two Exchange authors’ own published security tooling. SHPBL was combined with Casey Reid’s navi for three of them and with Isaac Conard’s targeted-nessus-scan for the other three — each pairing yielding software that neither side had, and neither side could have produced alone.
Agent evidence series
Composed entirely from SHPBL and CMPSBL capability intent, with no upstream project involved. Counterfactual Immune Forge is the one currently submitted; the other five are under author review before resubmission.
Under author review · MIT · 31 tests · 6,415 assertions
Which retrieved information influenced this decision, and how trustworthy was its supply chain?
Previous pull request #164 ↗Under author review · MIT · 60 tests · 158 assertions
What did the agent claim it did, and was that history altered?
Previous pull request #165 ↗Under author review · MIT · 71 tests · 127 assertions
Was this agent actually delegated this authority?
Previous pull request #166 ↗Under author review · MIT · 28 tests · 118 assertions
Has this agent's behaviour materially changed?
Previous pull request #167 ↗Under author review · MIT · 25 tests · 72 assertions
Did our synthetic evaluation data reappear somewhere it should not?
Previous pull request #168 ↗Submitted, awaiting review · MIT · 44 tests
Can a defense prove it improved against the exact scenario that beat it?
Tenable pull request #190 ↗Under author review · built on navi
Under author review · MIT · 33 tests · 180,000 invariant checks
Can this fix prove what it changed, and undo itself when verification fails?
Previous pull request #174 ↗Under author review · MIT · 25 tests · 120,000 invariant checks
Patch it or isolate it — which future is actually safer?
Previous pull request #175 ↗Under author review · MIT · 25 tests · 120,000 invariant checks
What is overdue, what fits in the window, and what did we fail to schedule?
Previous pull request #176 ↗Under author review · built on targeted-nessus-scan
Under author review · MIT · 29 tests · 210,000 invariant checks
What is the smallest scan that could disprove our worst theory about the fleet?
Previous pull request #171 ↗Under author review · MIT · 38 tests · 210,000 invariant checks
Which findings are one condition, and which neighbours will show it next?
Previous pull request #172 ↗Under author review · MIT · 39 tests · 420,000 invariant checks
Is this scan policy the best one, or just the one we inherited?
Previous pull request #173 ↗The public boundary
Each MIT grant covers only the named files in that one repository — its specification, or its independently written implementation and tests. It does not relicense SHPBL, CMPSBL, private harvest machinery, harvested bodies or provenance maps. Every implementation was written fresh from its published specification, cleared an exact-file review of its release surface, and received its own grant over that reviewed file set.