fix(5): decision-coverage gate parses <action> XML tag bodies for D-NN citations (#155)
* test(5): add failing test for decision IDs inside <objective>/<tasks>/<task>/<action> XML bodies Regression test for issue #5 — the translation gate (check.decision-coverage-plan) is blind to D-NN citations placed inside XML tag bodies by gsd-planner. Maintainer acceptance criteria (verbatim, issue #5): "Gate parses <action> tag bodies for decision ID citations; regression test with XML-tag plan body covers all decision IDs." Five new test cases added to the 'XML tag body citation parsing (issue #5)' suite: 1. RED: five decisions cited only in <objective>/<action> bodies → gate fails (before fix) 2. Non-canonical tag <comment> → must NOT count (negative control, passes) 3. Plain prose under undesignated heading → must NOT count (negative control, passes) 4. Self-closing <action/> → no crash, D-NN not covered (passes) 5. D-NN in <objective> body → should count (also fails before fix, GREEN after) Tests 1 and 5 are the load-bearing RED cases. All others are negative controls. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(5): extend extractPlanSections to scan <objective>/<tasks>/<task>/<action> for D-NN citations Closes #5. Root cause: extractPlanSections() in check-decision-coverage.ts collected only front-matter (must_haves/truths/objective) and body lines under designated markdown headings. The gsd-planner spec (agents/gsd-planner.md line 66) says 'Task actions reference the decision ID they implement (e.g., "per D-03")' and emits citations inside <action> tag bodies — a location the gate could not see. Fix: add extractXmlTagBodies() helper that matches the four canonical planner XML tags (<objective>, <tasks>, <task>, <action>) via a deliberately narrow regex (no XML parser library — D2 design decision). The helper output is appended to the designated string inside extractPlanSections(), making any D-NN citation inside those tag bodies count toward coverage. Maintainer acceptance criteria (verbatim, issue #5): "Gate parses <action> tag bodies for decision ID citations; regression test with XML-tag plan body covers all decision IDs." Self-closing tags (<action/>) are safely ignored — the capturing group does not match. Non-canonical tags (<comment>, <note>, etc.) are not in the alternation and are ignored by design. The CJS surface for check-decision-coverage.ts does NOT have a generator (gen-decisions.mjs covers decisions.ts, not this gate). No CJS artifact exists for this module. Per the generator framework established in PR #154 (ADR-3524), a CJS migration is a follow-up; this PR focuses on the TS fix. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * docs(5): clarify in gsd-planner.md that decision-coverage gate reads XML tag bodies Adds a parenthetical note to the existing self-check bullet (line 66) explaining which locations the gate scans so the planner's own guidance and the gate's behavior are explicitly aligned. Refs #5. No behavior change — documentation truthing only. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * chore(5): add changeset fragment for decision-coverage XML body fix Refs #5. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * docs(5): extract Interface Context for Executors into reference file to pass planner-decomposition gate gsd-planner.md was 49446 chars after the XML-tag clarification added in this PR, exceeding the 48K threshold enforced by tests/planner-decomposition.test.cjs. Extracted the "Interface Context for Executors" section (~2137 chars) into get-shit-done/references/planner-interface-context.md, leaving a one-line pointer in gsd-planner.md. New normalized size: 47310 chars (1842 chars under threshold). Refs #5 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(5): register planner-interface-context.md in INVENTORY.md and manifest - Bump References headline from 61 to 62 to match filesystem count - Add planner-interface-context.md row in Modular Planner Decomposition table - Update footnote from 61 to 62 top-level references - Regenerate docs/INVENTORY-MANIFEST.json via gen-inventory-manifest.cjs --write Fixes inventory-counts and inventory-manifest-sync CI failures caused by the extraction commit (32e8950f) adding a new reference file without updating the inventory artefacts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Tom Boucher
5
.changeset/5-decision-coverage-xml-bodies.md
Normal file
5
.changeset/5-decision-coverage-xml-bodies.md
Normal file
@@ -0,0 +1,5 @@
|
||||
---
|
||||
type: Fixed
|
||||
pr: 5
|
||||
---
|
||||
**`check.decision-coverage-plan` gate now recognises D-NN citations inside `<objective>`, `<tasks>`, `<task>`, and `<action>` XML tag bodies** — `extractPlanSections()` previously searched only front-matter (`must_haves`, `truths`, `objective`) and body lines under designated markdown headings. The `gsd-planner` spec directs agents to cite decision IDs inside `<action>` bodies; those citations were invisible to the gate, causing plans that correctly followed the spec to report `passed: false` with all decisions uncovered. `extractXmlTagBodies()` now extracts inner text from the four canonical planner XML tags via a narrow regex (no parser library), and that text is appended to the `designated` search string. Self-closing tags and non-canonical tags are safely ignored. (#5)
|
||||
@@ -64,6 +64,7 @@ The orchestrator provides user decisions in `<user_decisions>` tags from `/gsd:d
|
||||
**Self-check before returning:** For each plan, verify:
|
||||
- [ ] Every locked decision (D-01, D-02, etc.) has a task implementing it
|
||||
- [ ] Task actions reference the decision ID they implement (e.g., "per D-03")
|
||||
(The decision-coverage gate `check.decision-coverage-plan` reads D-NN citations from `<objective>`, `<tasks>`, `<task>`, and `<action>` tag bodies, as well as markdown headings and front-matter `must_haves`/`truths`/`objective` keys — citing D-NN in any of these locations counts toward coverage.)
|
||||
- [ ] No task implements a deferred idea
|
||||
- [ ] Discretion areas are handled reasonably
|
||||
|
||||
@@ -519,66 +520,7 @@ Wave numbers are pre-computed during planning. Execute-phase reads `wave` direct
|
||||
|
||||
## Interface Context for Executors
|
||||
|
||||
**Key insight:** "The difference between handing a contractor blueprints versus telling them 'build me a house.'"
|
||||
|
||||
When creating plans that depend on existing code or create new interfaces consumed by other plans:
|
||||
|
||||
### For plans that USE existing code:
|
||||
After determining `files_modified`, extract the key interfaces/types/exports from the codebase that executors will need:
|
||||
|
||||
```bash
|
||||
# Extract type definitions, interfaces, and exports from relevant files
|
||||
grep -n "export\\|interface\\|type\\|class\\|function" {relevant_source_files} 2>/dev/null | head -50
|
||||
```
|
||||
|
||||
Embed these in the plan's `<context>` section as an `<interfaces>` block:
|
||||
|
||||
```xml
|
||||
<interfaces>
|
||||
<!-- Key types and contracts the executor needs. Extracted from codebase. -->
|
||||
<!-- Executor should use these directly — no codebase exploration needed. -->
|
||||
|
||||
From src/types/user.ts:
|
||||
```typescript
|
||||
export interface User {
|
||||
id: string;
|
||||
email: string;
|
||||
name: string;
|
||||
createdAt: Date;
|
||||
}
|
||||
```
|
||||
|
||||
From src/api/auth.ts:
|
||||
```typescript
|
||||
export function validateToken(token: string): Promise<User | null>;
|
||||
export function createSession(user: User): Promise<SessionToken>;
|
||||
```
|
||||
</interfaces>
|
||||
```
|
||||
|
||||
### For plans that CREATE new interfaces:
|
||||
If this plan creates types/interfaces that later plans depend on, include a "Wave 0" skeleton step:
|
||||
|
||||
```xml
|
||||
<task type="auto">
|
||||
<name>Task 0: Write interface contracts</name>
|
||||
<files>src/types/newFeature.ts</files>
|
||||
<action>Create type definitions that downstream plans will implement against. These are the contracts — implementation comes in later tasks.</action>
|
||||
<verify>File exists with exported types, no implementation</verify>
|
||||
<done>Interface file committed, types exported</done>
|
||||
</task>
|
||||
```
|
||||
|
||||
### When to include interfaces:
|
||||
- Plan touches files that import from other modules → extract those module's exports
|
||||
- Plan creates a new API endpoint → extract the request/response types
|
||||
- Plan modifies a component → extract its props interface
|
||||
- Plan depends on a previous plan's output → extract the types from that plan's files_modified
|
||||
|
||||
### When to skip:
|
||||
- Plan is self-contained (creates everything from scratch, no imports)
|
||||
- Plan is pure configuration (no code interfaces involved)
|
||||
- Level 0 discovery (all patterns already established)
|
||||
See `get-shit-done/references/planner-interface-context.md` for the full interface extraction guide.
|
||||
|
||||
## Context Section Rules
|
||||
|
||||
|
||||
@@ -226,6 +226,7 @@
|
||||
"planner-gap-closure.md",
|
||||
"planner-graphify-auto-update.md",
|
||||
"planner-human-verify-mode.md",
|
||||
"planner-interface-context.md",
|
||||
"planner-mvp-mode.md",
|
||||
"planner-reviews.md",
|
||||
"planner-revision.md",
|
||||
|
||||
@@ -262,7 +262,7 @@ Full roster at `get-shit-done/workflows/*.md`. Workflows are thin orchestrators
|
||||
|
||||
---
|
||||
|
||||
## References (61 shipped)
|
||||
## References (62 shipped)
|
||||
|
||||
Full roster at `get-shit-done/references/*.md`. References are shared knowledge documents that workflows and agents `@-reference`. The groupings below match [`docs/ARCHITECTURE.md`](ARCHITECTURE.md#references-get-shit-donereferencesmd) — core, workflow, thinking-model clusters, and the modular planner decomposition.
|
||||
|
||||
@@ -353,11 +353,12 @@ The `gsd-planner` agent is decomposed into a core agent plus reference modules t
|
||||
| `planner-mvp-mode.md` | Vertical-slice planning rules for MVP mode. |
|
||||
| `planner-human-verify-mode.md` | Rules for `workflow.human_verify_mode = end-of-phase`: suppress `checkpoint:human-verify` task emission and route deferred items via `<verify><human-check>`. |
|
||||
| `planner-graphify-auto-update.md` | How `load_graph_context` surfaces `.last-build-status.json` auto-update state (running / failed / stale head) alongside the existing staleness annotation. Opt-in via `graphify.auto_update` (#3347). |
|
||||
| `planner-interface-context.md` | Interface context rules for executors — how to extract key interfaces/types/exports from existing code and document new interfaces that downstream plans will consume. |
|
||||
| `skeleton-template.md` | SKELETON.md template emitted for new-project Walking Skeleton (Phase 1 + `--mvp`). |
|
||||
| `user-story-template.md` | User story format for MVP planning — "As a / I want to / So that" structured fields. |
|
||||
| `spidr-splitting.md` | SPIDR splitting decomposition rules for handling large user stories in MVP mode. |
|
||||
|
||||
> **Subdirectory:** `get-shit-done/references/few-shot-examples/` contains additional few-shot examples (`plan-checker.md`, `verifier.md`) that are referenced from specific agents. These are not counted in the 61 top-level references.
|
||||
> **Subdirectory:** `get-shit-done/references/few-shot-examples/` contains additional few-shot examples (`plan-checker.md`, `verifier.md`) that are referenced from specific agents. These are not counted in the 62 top-level references.
|
||||
|
||||
---
|
||||
|
||||
|
||||
62
get-shit-done/references/planner-interface-context.md
Normal file
62
get-shit-done/references/planner-interface-context.md
Normal file
@@ -0,0 +1,62 @@
|
||||
# Interface Context for Executors
|
||||
|
||||
**Key insight:** "The difference between handing a contractor blueprints versus telling them 'build me a house.'"
|
||||
|
||||
When creating plans that depend on existing code or create new interfaces consumed by other plans:
|
||||
|
||||
## For plans that USE existing code:
|
||||
After determining `files_modified`, extract the key interfaces/types/exports from the codebase that executors will need:
|
||||
|
||||
```bash
|
||||
# Extract type definitions, interfaces, and exports from relevant files
|
||||
grep -n "export\\|interface\\|type\\|class\\|function" {relevant_source_files} 2>/dev/null | head -50
|
||||
```
|
||||
|
||||
Embed these in the plan's `<context>` section as an `<interfaces>` block:
|
||||
|
||||
```xml
|
||||
<interfaces>
|
||||
<!-- Key types and contracts the executor needs. Extracted from codebase. -->
|
||||
<!-- Executor should use these directly — no codebase exploration needed. -->
|
||||
|
||||
From src/types/user.ts:
|
||||
```typescript
|
||||
export interface User {
|
||||
id: string;
|
||||
email: string;
|
||||
name: string;
|
||||
createdAt: Date;
|
||||
}
|
||||
```
|
||||
|
||||
From src/api/auth.ts:
|
||||
```typescript
|
||||
export function validateToken(token: string): Promise<User | null>;
|
||||
export function createSession(user: User): Promise<SessionToken>;
|
||||
```
|
||||
</interfaces>
|
||||
```
|
||||
|
||||
## For plans that CREATE new interfaces:
|
||||
If this plan creates types/interfaces that later plans depend on, include a "Wave 0" skeleton step:
|
||||
|
||||
```xml
|
||||
<task type="auto">
|
||||
<name>Task 0: Write interface contracts</name>
|
||||
<files>src/types/newFeature.ts</files>
|
||||
<action>Create type definitions that downstream plans will implement against. These are the contracts — implementation comes in later tasks.</action>
|
||||
<verify>File exists with exported types, no implementation</verify>
|
||||
<done>Interface file committed, types exported</done>
|
||||
</task>
|
||||
```
|
||||
|
||||
## When to include interfaces:
|
||||
- Plan touches files that import from other modules → extract those module's exports
|
||||
- Plan creates a new API endpoint → extract the request/response types
|
||||
- Plan modifies a component → extract its props interface
|
||||
- Plan depends on a previous plan's output → extract the types from that plan's files_modified
|
||||
|
||||
## When to skip:
|
||||
- Plan is self-contained (creates everything from scratch, no imports)
|
||||
- Plan is pure configuration (no code interfaces involved)
|
||||
- Level 0 discovery (all patterns already established)
|
||||
@@ -517,3 +517,197 @@ describe('config-type validation (review F16)', () => {
|
||||
expect(warnings.some((w) => /context_coverage_gate.*invalid type/.test(w))).toBe(true);
|
||||
});
|
||||
});
|
||||
|
||||
// ─── XML tag body parsing (issue #5) ──────────────────────────────────────
|
||||
|
||||
/**
|
||||
* Regression tests for issue #5: the translation gate must parse decision IDs
|
||||
* from <objective>, <tasks>, <task>, and <action> tag bodies.
|
||||
*
|
||||
* The gsd-planner spec (agents/gsd-planner.md line 66) says:
|
||||
* "Task actions reference the decision ID they implement (e.g., 'per D-03')"
|
||||
* Plans produced by that spec put D-NN citations inside <action> bodies,
|
||||
* which the gate was blind to before this fix.
|
||||
*
|
||||
* Maintainer acceptance criteria (verbatim from issue #5):
|
||||
* "Gate parses <action> tag bodies for decision ID citations; regression test
|
||||
* with XML-tag plan body covers all decision IDs."
|
||||
*/
|
||||
describe('XML tag body citation parsing (issue #5)', () => {
|
||||
/** Five locked decisions, D-01..D-05. */
|
||||
const FIVE_DECISIONS = `<decisions>
|
||||
### Architecture
|
||||
- **D-01:** Use strict TypeScript everywhere
|
||||
- **D-02:** Prefer functional composition over class inheritance
|
||||
- **D-03:** All public APIs must have JSDoc
|
||||
- **D-04:** Errors are typed, never thrown as plain strings
|
||||
- **D-05:** Async functions always return explicit Promises
|
||||
</decisions>`;
|
||||
|
||||
/**
|
||||
* Plan whose D-NN citations are ONLY inside XML tag bodies —
|
||||
* no front-matter must_haves/truths/objective entries.
|
||||
*/
|
||||
const XML_PLAN = `---
|
||||
phase: 17
|
||||
plan: 1
|
||||
type: implementation
|
||||
wave: 1
|
||||
depends_on: []
|
||||
files_modified: []
|
||||
autonomous: true
|
||||
must_haves:
|
||||
- thing-a
|
||||
objective: |
|
||||
Build the foo.
|
||||
---
|
||||
|
||||
<objective>Implement foo per D-01 and D-02 design</objective>
|
||||
|
||||
<tasks>
|
||||
<task>
|
||||
<action>Apply D-03 mitigation to the bar module</action>
|
||||
</task>
|
||||
<task>
|
||||
<action>Refactor per D-04 — ensures D-05 invariant</action>
|
||||
</task>
|
||||
</tasks>
|
||||
`;
|
||||
|
||||
it('RED: gate is blind to D-NN citations inside XML tag bodies (before fix)', async () => {
|
||||
// This test is expected to FAIL on origin/main (gate returns covered: 0).
|
||||
// After fix it should pass (covered: 5).
|
||||
await setupPhase(FIVE_DECISIONS, { '17-01-PLAN.md': XML_PLAN });
|
||||
|
||||
const result = await checkDecisionCoveragePlan([phaseDir, contextPath], tmp);
|
||||
|
||||
// GREEN assertion (what we want AFTER the fix):
|
||||
expect(result.data.passed).toBe(true);
|
||||
expect(result.data.covered).toBe(5);
|
||||
expect(result.data.uncovered).toEqual([]);
|
||||
expect(result.data.total).toBe(5);
|
||||
});
|
||||
|
||||
it('does NOT count D-NN inside a non-canonical XML tag like <comment>', async () => {
|
||||
// <comment> is not one of the four canonical tags; must not count.
|
||||
const planWithComment = `---
|
||||
phase: 17
|
||||
plan: 1
|
||||
type: implementation
|
||||
wave: 1
|
||||
depends_on: []
|
||||
files_modified: []
|
||||
autonomous: true
|
||||
must_haves:
|
||||
- thing-a
|
||||
---
|
||||
|
||||
<comment>D-01 not implemented yet</comment>
|
||||
`;
|
||||
await setupPhase(
|
||||
`<decisions>
|
||||
### Cat
|
||||
- **D-01:** A trackable decision six words or more
|
||||
</decisions>`,
|
||||
{ '17-01-PLAN.md': planWithComment },
|
||||
);
|
||||
|
||||
const result = await checkDecisionCoveragePlan([phaseDir, contextPath], tmp);
|
||||
expect(result.data.passed).toBe(false);
|
||||
expect(result.data.uncovered.map((u: { id: string }) => u.id)).toContain('D-01');
|
||||
});
|
||||
|
||||
it('does NOT count D-NN mentioned in regular prose outside any XML tag', async () => {
|
||||
// Prose outside designated headings and XML tags must not count.
|
||||
const planWithProse = `---
|
||||
phase: 17
|
||||
plan: 1
|
||||
type: implementation
|
||||
wave: 1
|
||||
depends_on: []
|
||||
files_modified: []
|
||||
autonomous: true
|
||||
must_haves:
|
||||
- thing-a
|
||||
---
|
||||
|
||||
This section mentions D-01 in plain prose under an undesignated heading.
|
||||
|
||||
## Design Notes
|
||||
|
||||
D-01 appears here too but this heading is not designated.
|
||||
`;
|
||||
await setupPhase(
|
||||
`<decisions>
|
||||
### Cat
|
||||
- **D-01:** A trackable decision six words or more
|
||||
</decisions>`,
|
||||
{ '17-01-PLAN.md': planWithProse },
|
||||
);
|
||||
|
||||
const result = await checkDecisionCoveragePlan([phaseDir, contextPath], tmp);
|
||||
expect(result.data.passed).toBe(false);
|
||||
expect(result.data.uncovered.map((u: { id: string }) => u.id)).toContain('D-01');
|
||||
});
|
||||
|
||||
it('does not crash on self-closing <action/> tags', async () => {
|
||||
const planWithSelfClosing = `---
|
||||
phase: 17
|
||||
plan: 1
|
||||
type: implementation
|
||||
wave: 1
|
||||
depends_on: []
|
||||
files_modified: []
|
||||
autonomous: true
|
||||
must_haves:
|
||||
- thing-a
|
||||
---
|
||||
|
||||
<tasks>
|
||||
<task>
|
||||
<action/>
|
||||
</task>
|
||||
</tasks>
|
||||
`;
|
||||
await setupPhase(
|
||||
`<decisions>
|
||||
### Cat
|
||||
- **D-01:** A trackable decision six words or more
|
||||
</decisions>`,
|
||||
{ '17-01-PLAN.md': planWithSelfClosing },
|
||||
);
|
||||
|
||||
// Should not throw; D-01 is uncovered (no citation in self-closing tag)
|
||||
const result = await checkDecisionCoveragePlan([phaseDir, contextPath], tmp);
|
||||
expect(result.data.passed).toBe(false);
|
||||
expect(result.data.uncovered.map((u: { id: string }) => u.id)).toContain('D-01');
|
||||
});
|
||||
|
||||
it('counts D-NN in <objective> tag body', async () => {
|
||||
const planWithObjective = `---
|
||||
phase: 17
|
||||
plan: 1
|
||||
type: implementation
|
||||
wave: 1
|
||||
depends_on: []
|
||||
files_modified: []
|
||||
autonomous: true
|
||||
must_haves:
|
||||
- thing-a
|
||||
---
|
||||
|
||||
<objective>Implement D-01 as the core type system</objective>
|
||||
`;
|
||||
await setupPhase(
|
||||
`<decisions>
|
||||
### Cat
|
||||
- **D-01:** A trackable decision six words or more
|
||||
</decisions>`,
|
||||
{ '17-01-PLAN.md': planWithObjective },
|
||||
);
|
||||
|
||||
const result = await checkDecisionCoveragePlan([phaseDir, contextPath], tmp);
|
||||
expect(result.data.passed).toBe(true);
|
||||
expect(result.data.covered).toBe(1);
|
||||
});
|
||||
});
|
||||
|
||||
@@ -157,6 +157,21 @@ interface PlanSections {
|
||||
|
||||
const DESIGNATED_HEADINGS_RE = /^#{1,6}\s+(?:must[_ ]haves?|truths?|tasks?|objective)\b/i;
|
||||
|
||||
/**
|
||||
* Extracts text from the four canonical XML tags that gsd-planner emits for
|
||||
* decision citations: <objective>, <tasks>, <task>, <action>.
|
||||
*
|
||||
* Uses a deliberately narrow regex (D2) — no XML parser library — because the
|
||||
* planner's XML convention is a project-internal convention, not formal XML.
|
||||
* Self-closing tags (e.g., <action/>) are harmlessly skipped (no capture group
|
||||
* match). Non-canonical tags (e.g., <comment>) are NOT matched.
|
||||
*
|
||||
* Reference: agents/gsd-planner.md line 66:
|
||||
* "Task actions reference the decision ID they implement (e.g., 'per D-03')"
|
||||
*/
|
||||
const XML_DECISION_TAGS_RE =
|
||||
/<(?:objective|tasks?|action)(?:\s[^>]*)?>([\s\S]*?)<\/(?:objective|tasks?|action)>/gi;
|
||||
|
||||
/** Strip HTML comments AND fenced code blocks from `text`. */
|
||||
function stripCommentsAndFences(text: string): string {
|
||||
return text
|
||||
@@ -165,6 +180,19 @@ function stripCommentsAndFences(text: string): string {
|
||||
.replace(/~~~[\s\S]*?~~~/g, ' ');
|
||||
}
|
||||
|
||||
/**
|
||||
* Extract the inner text of all canonical XML tag bodies from `text`.
|
||||
* Returns a concatenated string of all matched inner bodies, or '' if none.
|
||||
* Called on the cleaned (comments+fences stripped) plan body.
|
||||
*/
|
||||
function extractXmlTagBodies(text: string): string {
|
||||
const parts: string[] = [];
|
||||
for (const m of text.matchAll(XML_DECISION_TAGS_RE)) {
|
||||
if (m[1]) parts.push(m[1]);
|
||||
}
|
||||
return parts.join('\n');
|
||||
}
|
||||
|
||||
/** Extract a YAML block scalar (key followed by indented continuation lines). */
|
||||
function extractYamlBlock(frontmatter: string, key: string): string {
|
||||
const re = new RegExp(`^${key}\\s*:(.*)$`, 'm');
|
||||
@@ -214,7 +242,14 @@ function extractPlanSections(planContent: string): PlanSections {
|
||||
if (inDesignated) bodyParts.push(line);
|
||||
}
|
||||
|
||||
return { designated: [...fmParts, bodyParts.join('\n')].join('\n\n') };
|
||||
// Also include the inner text of canonical XML tag bodies (<objective>, <tasks>,
|
||||
// <task>, <action>). The planner spec (agents/gsd-planner.md line 66) directs
|
||||
// agents to cite D-NN inside <action> bodies; the gate must honour those citations.
|
||||
// extractXmlTagBodies is called on the full cleaned content (not just the body
|
||||
// portion) so that <objective> blocks at the top of the document are also caught.
|
||||
const xmlParts = extractXmlTagBodies(cleaned);
|
||||
|
||||
return { designated: [...fmParts, bodyParts.join('\n'), xmlParts].join('\n\n') };
|
||||
}
|
||||
|
||||
async function loadPlanSections(phaseDir: string): Promise<PlanSections[]> {
|
||||
|
||||
Reference in New Issue
Block a user