turkananation

Maintain pqforge documentation in lockstep with the code, with zero hallucinated APIs, commands, numbers, or links. Use whenever editing README.md, CHANGELOG.md, anything under doc/ or wiki/, the visibility manifest (tool/visibility/visibility_manifest.json), the generator, or any generated AI-discovery file (llms.txt, llms-full.txt, identity.json, faq-ai.txt, developer-ai.txt, ai.txt, robots*.txt, site/**, .github/copilot-instructions.md, .github/instructions/**, .cursor/rules/**, .windsurfrules). Enforces the verify-against-code protocol, the generated-vs-hand-maintained split, the claim boundary, the main-branch link rule, and the regenerate-and-validate checklist.

Implement pqforge library, CLI, codec, recipe, streaming, hybrid, custody, test, or example features. Use for behavior changes under lib/**, bin/**, example/**, or test/** and for cross-cutting feature refactors.

Prepare and verify pqforge package releases. Use for version changes, changelog release notes, pub.dev archive checks, release branches, tags, GitHub binaries, or pub.dev publication.

Maintain pqforge documentation from live code evidence. Use when editing README.md, CHANGELOG.md, CLAUDE.md, AGENTS.md, doc/**, wiki/**, the visibility manifest or generator, generated AI-discovery/editor files, or site/**. Also use when verifying API names, CLI commands, flags, performance numbers, package claims, links, or generated-document consistency.

Implement pqforge features and fixes across the library, CLI, codecs, recipes, streaming, hybrid cryptography, key custody, tests, and examples. Use for any behavior change under lib/**, bin/**, example/**, or test/**, including new APIs, commands, flags, formats, cipher/provider behavior, performance changes, and cross-cutting refactors.

Prepare, verify, and audit pqforge releases. Use when changing the package version, updating release notes, preparing a release branch or tag, checking the pub.dev archive, validating generated version/visibility files, building release binaries, or publishing to pub.dev/GitHub.

Evidence-scoped documentation workflow for the pqcrypto repository. Use when updating README.md, doc/ guides, test-data provenance, release roadmaps, interop docs, cookbook docs, generated-visibility manifests, or any claim about delivered ML-KEM, ML-DSA, SLH-DSA, FIPS 202, FIPS 180-4, OpenSSL/liboqs interop, validation evidence, package boundaries, or release readiness.

Project-level PQC multi-agent framework wrapper for Claude Code. Uses the canonical pqcrypto evidence-scoped manifest and docs for Cryptographic Architect, SecOps, Distinguished Engineer, and Client Integration workflows.

Project-level PQC multi-agent framework wrapper for Codex. Uses the canonical pqcrypto evidence-scoped manifest and docs for Cryptographic Architect, SecOps, Distinguished Engineer, and Client Integration workflows.

Project-level PQC multi-agent framework wrapper for Antigravity. Uses the canonical pqcrypto evidence-scoped manifest and docs for Cryptographic Architect, SecOps, Distinguished Engineer, and Client Integration workflows.

Convene a chosen conclave of personas for grounded, multi-perspective deliberation on a hard decision — code/architecture calls, strategic or institutional direction, red-team war-gaming, or risk review. Use whenever the user runs /conclave, asks to "war-game", "deliberate", "stress-test", "red-team", or "get The Sovereign Conclave on" a decision, or wants several opposed expert viewpoints to argue a problem before a recommendation. Produces a written verdict grounded in supplied evidence; advises only — the human decides.