speckit-evidence-graph
Validate and render the task DAG; compute synthetic propagation.
- License 许可
- MIT
- 最近更新
- 2026-06-16
- Stars
- ★ 2
- Forks
- 0
01 前言 Front Matter
- name
- speckit-evidence-graph
- metadata
- { "author": "github-spec-kit", "source": "evidence:commands/speckit.evidence.graph.md" }
- description
- Validate and render the task DAG; compute synthetic propagation.
- compatibility
- Requires spec-kit project structure with .specify/ directory
03 SKILL.md 正文
/speckit.evidence.graph
Parse specs/<feature>/tasks.md and specs/<feature>/tasks.deps.yml,
validate the graph (acyclic, no dangling refs, every id present in both
files), validate task skillist metadata and mirrors, and render
readiness/task-graph.json + readiness/task-graph.md.
How to invoke
./fake.sh build -t EvidenceGraph
The graph computes in-process in compiled F#
(FS.Skia.UI.Build.Evidence.Engine.runGraph); there is no Python or shell audit
runner. The gate reads tasks.md / tasks.deps.yml / readiness/ at the
build.fsx interpreter edge and writes readiness/task-graph.json +
readiness/task-graph.md.
When to run
- Right after
/speckit.tasks— confirms the initial DAG is well-formed before implementation begins. - After every status change during
/speckit.implement— refreshes[S*]propagation cheaply. - Automatically as the
before_implementhook (declared in the evidence extension'sextension.yml) — refuses to start implement on a broken graph.
What it validates
- Every
Tnnnintasks.mdhas a matching key intasks.deps.yml. - Every
Tnnnintasks.deps.ymlhas a matching task line intasks.md. - Every dep reference resolves to a known
Tnnn. - The graph is acyclic.
- No task depends on itself.
- Every task has object-form metadata with
depsandskillist. - Every
tasks.mdtask line mirrors the structuredskillist. - Declared skill ids resolve to exactly one readable local capability skill.
- Obvious capability omissions and invalid multi-skill prerequisite ordering are readiness failures.
What it computes
Effective status per task, under the propagation rule:
effective(T) =
synthetic if declared(T) == synthetic
auto-synthetic if declared(T) == done AND any dep is (auto-)synthetic
declared(T) otherwise
Phase-checkpoint edges are auto-injected (every task in Phase N+1 gets an
implicit edge to the last foundation task of Phase N). These do not appear in
tasks.deps.yml and need not be written by hand.
On failure
The script exits non-zero and writes the errors into task-graph.md's
verdict block. Do not proceed with /speckit.implement until the graph is
clean. Common failure modes and their fixes:
- Dangling ref —
tasks.deps.ymlreferencesTnnnthat isn't intasks.md. Add the task line or remove the ref. - Orphaned key —
tasks.deps.ymlhas a key forTnnnthat isn't intasks.md. Remove the key or add the task line. - Cycle — a set of tasks transitively depend on each other. The error message names the cycle path. Break it by removing one edge.
- Duplicate task id — the same
Tnnnappears twice intasks.md. Renumber one. - Missing or invalid
skillist— migrate or regenerate the task list so every task has structuredskillistmetadata and a matching visible mirror. - Unresolved skill — fix the skill id, restore the missing
SKILL.md, or resolve ambiguous duplicated skill ids before implementation.
Output
specs/<FEATURE_ID>/readiness/task-graph.json— structured state.specs/<FEATURE_ID>/readiness/task-graph.md— mermaid diagram, ASCII view, status counts, propagation report with root-cause annotations.
Commit both files alongside the feature's other artifacts.
Authoritative status region (spec 037, US2)
Machine-readable status values are read only from a fenced code block whose
info string is exactly audit-status. Prose, markdown bullets, and any other
fenced block are never read as status, so a blocker term inside explanatory text
or a negation cannot raise a false block (FR-004, FR-005).
Deterministic resolution rule:
- First region wins — the first
audit-statusregion that declares a key provides its authoritative value. - Duplicate key within the region is a parse error — never silent last-wins.
- Prose never wins — a key in prose/bullets/other blocks is ignored.
- Malformed entry (missing
=, empty key) is a parse error — never silently treated as passing or failing.
Blocking is structured, not substring (FR-006): the audit blocks on explicit
violating values (exact-package-match not in {true,yes},
package-resolution=nu1603, taskbar-only=true, or taskbar-entry=true with
window-visible=false) — never on substring presence of taskbar-only /
mismatch / nu1603 in text. Scanner:
.specify/extensions/evidence/scripts/python/audit-status-scan.py.
Sequential FAKE Commands
FAKE-backed commands (./fake.sh, fake.cmd, or dotnet fake) share
repository .fake state and are not safe to run concurrently. Non-FAKE graph
file reads may run in parallel when they do not depend on .fake, but
multiple FAKE-backed targets must run sequentially:
./fake.sh build -t EvidenceGraph./fake.sh build -t EvidenceAudit
04 同仓库更多 Skills
fs-skia-ui-widgets
Generated product guidance for Skia-rendered FS.Skia.UI Controls, rich text, chart controls, graph controls, DataGrid, and custom wrappers.
fs-skia-testing
Assert generated-product expectations and evidence in a governed FS.Skia.UI product.
fs-skia-skiaviewer
Wire a generated FS.Skia.UI product to the desktop viewer host.
fs-skia-scene
Build pure scene descriptions in a generated FS.Skia.UI product.
fs-skia-keyboard-input
Map keyboard input to product commands in a generated FS.Skia.UI product.
fs-skia-elmish
Drive a generated FS.Skia.UI product through the pure Elmish adapter.