enhance(#1577): WebFetch/WebSearch injection isolation + opt-in blocking (#1585)

* fix(#1577): isolate WebFetch/WebSearch ingress + opt-in injection blocking

Split A of #1573 (security-critical). Scans WebFetch/WebSearch output (the
largest untrusted channel) in gsd-read-injection-scanner; shared
untrusted-input-boundary reference @-included by the 8 ingest agents
(randomized per-wrap delimiters, in-prompt self-scan guard, task-anchoring);
opt-in security.injection_blocking (default advisory — non-breaking).

arXiv: 2506.05739 (PPA), 2507.15219 (PromptArmor), 2504.20472 (Referencing), 2503.00061 (defense-in-depth).

* fix(#1577): address review — honest blocking docs, config key, ADR, property test, revert localized

- A1: rewrote the opt-in-blocking doc + Security changeset honestly — the PostToolUse hook is a
  circuit-breaker (halts the agent's next step), NOT a redactor; it does not scrub content already
  in the transcript. The prompt-level data/instruction boundary is the primary control.
- A2: registered security.injection_blocking in the config schema + defaults manifests (default
  false) + an e2e config-roundtrip test; the dotted setter writes the nested shape the hook reads.
- A3: reverted the 4 hand-edited localized security-model.md (canonical EN only, per convention).
- A5: ADR-1577 (untrusted-input boundary + opt-in blocking; redaction-vs-circuit-breaker rationale).
- A6: property test — scanner never crashes / only emits valid JSON on unicode/large/malformed input.
- Also: inventory (untrusted-input-boundary.md) + agent-size baseline (8 ingest agents) +
  drift-guard matcher update (Read -> Read|WebFetch|WebSearch). A7 (content<20 early-exit) left as
  the noted pre-existing follow-up.

* fix(#1577): allowlist untrusted-input-boundary.md in injection-scan CI gate

The new reference quotes injection phrases ('ignore previous instructions',
'you are now…') as examples agents must NOT comply with, tripping the repo's
own prompt-injection-scan.sh diff gate (the standalone 'security' CI job, red
on HEAD). Allowlist it alongside the other security docs (security-model.md,
TEST-EXAMPLES.md) that legitimately demonstrate injection patterns. The JS
scanner test doesn't scan references/, so only the shell gate needed it.

Verified: scan --diff origin/next -> 0 findings; scanner JS test 15/15.

* fix(#1577): cover AC #2's gsd-ui-researcher + gsd-assumptions-analyzer

trek-e Major 1: the @-included set dropped two AC #2 agents. Restore them so
no named web-ingress agent is uncovered, keeping the two justified additions
(gsd-ai-researcher, gsd-domain-researcher). Final set = AC's 8 + 2 = 10.
 - gsd-ui-researcher carries the full WebSearch/WebFetch + MCP-fetch toolset.
 - gsd-assumptions-analyzer reads 5-15 codebase source files (external/source-
   document ingress per the boundary), though it has no web tools.
INGEST_AGENTS in the isolation test now asserts all 10; size baselines
regenerated (+60 bytes each, both well under the DEFAULT cap); changeset
reworded 8 -> 10.

Verified: untrusted-input-isolation 14/14; agent-size-budget 39/39.

* docs(#1577): document security.injection_blocking + boundary seam

trek-e Major 2 + Minor:
 - docs/CONFIGURATION.md: add the top-level security.injection_blocking key to
   the Full Schema and a Security Settings subsection, distinguishing it from
   the workflow.security_* namespace; honest circuit-breaker-not-redactor
   framing matching ADR-1577 / security-model.
 - CONTEXT.md: add the 'Untrusted-input boundary' seam glossary entry.

Verified: lint:docs ok; config-field-docs + contributor-standards green.

* test(#1577): make read-injection property test git-text, not binary

trek-e nit (and more): the file embedded a raw U+FFFF AND a raw NUL byte as
degenerate-edge inputs. The NUL is what actually made git classify it binary
(git binary = NUL in first 8K). Replace both with text-safe escapes that keep
the identical runtime values: '\\x00' and String.fromCodePoint(0xFFFF). File
now diffs/blames line-by-line.

Verified: property test 2/2; no NUL/raw-noncharacter bytes remain.

* docs(#1577): align untrusted boundary docs

Name all 10 ingress agents in INVENTORY/security-model and allowlist the intentional read-injection property corpus for the prompt-injection scanner.

* docs(#1577): align ADR ingest agent count

Update ADR-1577 from 8 to 10 ingest agents so it matches the actual boundary include set and the rest of the docs.

---------

Co-authored-by: Tom Boucher <trekkie@nomorestars.com>
This commit is contained in:
Alex V.
2026-06-25 00:07:23 +03:00
committed by GitHub
parent 3870fafe74
commit a63684c222
29 changed files with 536 additions and 52 deletions

View File

@@ -17,6 +17,8 @@ Spawned by `discuss-phase` via `Task()`. You do NOT present output directly to t
- Return structured markdown output for the main agent to synthesize
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<documentation_lookup>
@~/.claude/gsd-core/references/research-documentation-lookup.md
</documentation_lookup>

View File

@@ -16,6 +16,8 @@ You are a GSD AI researcher. Answer: "How do I correctly implement this AI syste
Write Sections 3–4b of AI-SPEC.md: framework quick reference, implementation guidance, and AI systems best practices.
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<documentation_lookup>
@~/.claude/gsd-core/references/research-documentation-lookup.md
</documentation_lookup>

View File

@@ -18,6 +18,8 @@ Spawned by `discuss-phase-assumptions` via `Task()`. You do NOT present output d
- Flag topics where codebase analysis alone is insufficient (needs external research)
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<input>
Agent receives via prompt:

View File

@@ -18,6 +18,8 @@ You are a GSD doc classifier. You read ONE document and write a structured class
If the prompt contains a `<required_reading>` block, use the `Read` tool to load every file listed there before doing anything else. That is your primary context.
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<why_this_matters>
Your classification drives extraction. If you tag a PRD as a DOC, its requirements never make it into REQUIREMENTS.md. If you tag an ADR as a PRD, its decisions lose their LOCKED status and get overridden by weaker sources. Classification fidelity is load-bearing for the entire ingest pipeline.
</why_this_matters>

View File

@@ -20,6 +20,8 @@ You do NOT prompt the user. You do NOT write PROJECT.md, REQUIREMENTS.md, or ROA
If the prompt contains a `<required_reading>` block, load every file listed there first — especially `references/doc-conflict-engine.md` which defines your conflict report format.
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<why_this_matters>
You are the precedence-enforcing layer. Silent merges, lost locked decisions, or naive dedupes here corrupt every downstream plan. When in doubt, surface the conflict rather than pick.
</why_this_matters>

View File

@@ -16,6 +16,8 @@ You are a GSD domain researcher. Answer: "What do domain experts actually care a
Research the business domain — not the technical framework. Write Section 1b of AI-SPEC.md.
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<documentation_lookup>
@~/.claude/gsd-core/references/research-documentation-lookup.md
</documentation_lookup>

View File

@@ -35,6 +35,8 @@ Spawned by `/gsd:plan-phase` (integrated) or `/gsd:plan-phase --research-phase <
Claims tagged `[ASSUMED]` signal to the planner and discuss-phase that the information needs user confirmation before becoming a locked decision. Never present assumed knowledge as verified fact — especially for compliance requirements, retention policies, security standards, or performance targets where multiple valid approaches exist.
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<documentation_lookup>
@~/.claude/gsd-core/references/research-documentation-lookup.md
</documentation_lookup>

View File

@@ -32,6 +32,8 @@ Your files feed the roadmap:
**Be comprehensive but opinionated.** "Use X because Y" not "Options are X, Y, Z."
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<documentation_lookup>
@~/.claude/gsd-core/references/research-documentation-lookup.md
</documentation_lookup>

View File

@@ -32,6 +32,8 @@ If the prompt contains a `<required_reading>` block, you MUST use the `Read` too
- Commit ALL research files (researchers write but don't commit — you commit everything)
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<downstream_consumer>
Your SUMMARY.md is consumed by the gsd-roadmapper agent which uses it to:

View File

@@ -27,6 +27,8 @@ If the prompt contains a `<required_reading>` block, you MUST use the `Read` too
- Return structured result to orchestrator
</role>
@~/.claude/gsd-core/references/untrusted-input-boundary.md
<documentation_lookup>
@~/.claude/gsd-core/references/research-documentation-lookup.md
</documentation_lookup>