feat(host): prompt ingestion as a local provider (ADR 0006)

[macanderson]

What & why

The prompt's biggest un-disciplined input is the text a user pastes: a realistic turn mixes a log, a table, a directory reference, and the actual ask under one blob, and only the last is intent. Pasted whole it is re-sent verbatim every turn (no cache, no dedup), its cost is never accounted, nothing is content-addressed, and the model is handed material it must itself decide is mostly irrelevant.

This models the paste as a local, egress-free CGP provider — the ingestion-side dual of compose_context. It's host-side reference behavior; no envelope shape or SPEC.md change, using only the representation fields ADR 0005 added.

• intent → query.goal, verbatim (never rewritten — the one load-bearing UX guarantee)
• directory references → query.anchors (zero tokens; the graph provider resolves them)
• evidence (logs, tables, code, notes) → content-addressed frames, served compact by default with the full bytes rehydratable via a [full] re-query

The guarantee is not "zero wasted tokens" (relevance is only knowable downstream) but bounded default cost with lossless retrieval.

Honest by construction

• token_cost and the inline content_digest are recomputed for the exact representation emitted (§B3) — never carried across a representation flip
• provenance is derivation, not file, so it doesn't trip §F5 (a paste has no re-readable URI); the real hash lives in canonical_content_hash
• content-addressed + immutable ⇒ verify is exact (valid / stale / gone), and the same paste dedups to one frame
• local-only (egress: false, LocalOnly scope) ⇒ auto-permitted, no consent friction

resolve: true vs ADR 0004

Serving compact/reference requires resolve: true (representations_consistent). Unlike the upsert/subscribe flags ADR 0004 deleted, resolve has a live consistency rule and a working, tested in-process rehydration path today (a [full]-preference re-query returns the full bytes from the store). The ADR spells this out; the wire context/resolve method remains a later phase.

Wire conformance — and one corrective schema fix

Verifying "frames are wire-conformant" (not just Rust-invariant-conformant) surfaced a real, pre-existing schema/serializer mismatch: the JSON Schema listed provenance and relations as globally required on a ContextFrame, but the reference serializer omits both when empty (skip_serializing_if = Vec::is_empty) and no frame-validity check requires either. Every ingest frame has no graph edges, so it failed schema validation. Prior schema coverage only validated hand-authored example transcripts — never Rust-serialized frames — so the mismatch sat unexercised.

• schema: ContextFrame.required is now exactly id/kind/title/score/token_cost (what the serializer always emits); content stays governed per-representation by the existing allOf.
• Empirically confirmed with jsonschema: all three representations validate against schema/contextgraph-envelope.schema.json; existing examples still pass.

Tests

• contextgraph-host/src/ingest/tests.rs — segmentation, §B3 honesty across all three representations, representation invariants, SHA-256 known-answer vectors (incl. padding boundaries), content-addressed dedup, query budget/frame caps, verify
• contextgraph-host/tests/prompt_ingest.rs — end-to-end through the real Host: fan-out, budget audit, byte-stable compose, verify_frames
• contextgraph-conformance/tests/ingest_conformance.rs — real ingested frames (full/compact/reference) through check_frames + check_budget + representation_invariants + the schema's required-key set + NDJSON envelope round-trip (the guard that catches serialized frames drifting from the schema)

cargo test --workspace, cargo clippy --workspace --all-targets -- -D warnings, cargo fmt --all --check, and python3 schema/validate-examples.py all green. New direct dep: sha2 (host crate only; contextgraph-types stays zero-dep beyond serde).

Not in scope (additive follow-ups)

Stack-trace parsing and richer tabular inference behind the existing classify seam; the context/resolve wire method; making provenance non-required is left as-is (ingest frames always carry it).

[sourcery-ai]

Sorry @macanderson, you have reached your weekly rate limit of 500000 diff characters.

Please try again later or upgrade to continue using Sourcery

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
context-graph-protocol	Ready	Preview, Comment	Jul 23, 2026 7:35pm

[sourcery-ai]

Sorry @macanderson, you have reached your weekly rate limit of 500000 diff characters.

Please try again later or upgrade to continue using Sourcery

[macanderson]

Filed the remaining follow-ups so nothing is left implicit:

• Deepen prompt-ingest segmentation & distillation (ADR 0006 follow-ups) #53 — Deepen prompt-ingest segmentation & distillation (stack traces, richer tables, log-timestamp normalization, dup-line collapse, classification-precedence edge cases). All additive, non-normative.
• Schema required must match the reference serializer's omit-when-empty fields — audit every wire type, and validate serialized reference values #54 — Schema required must match the reference serializer's omit-when-empty fields: audit every wire type + add a serialize-and-validate conformance test. This is the systemic version of the schema fix in this PR (which corrected only ContextFrame).
• Prompt-ingest: a docs page and a runnable example #55 — Prompt-ingest docs page + runnable example.

context/resolve is already tracked by #50; I added a comment there noting this PR's IngestProvider as a concrete compact/reference consumer with a working in-process rehydration path (and its per-representation B3 handling), rather than opening a duplicate.

Correction to the PR description's "Not in scope" note: this PR removed both provenance and relations from the frame's required list (the reference serializer omits both when empty), not just relations — ingest frames always carry provenance, but the schema no longer requires it, matching the serializer. #54 covers auditing the same pattern across the other wire types.