sdd-flow 1.3.3 → 2.0.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (70) hide show
  1. package/CHANGELOG.md +33 -0
  2. package/CONTRIBUTING.md +25 -0
  3. package/LICENSE +21 -21
  4. package/README.md +113 -168
  5. package/README.pt-BR.md +100 -0
  6. package/bin/cli.js +162 -424
  7. package/docs/architecture.md +91 -0
  8. package/docs/distribution.md +64 -0
  9. package/docs/formats.md +212 -0
  10. package/docs/local-testing.pt-BR.md +195 -0
  11. package/docs/methodology.pt-BR.md +133 -0
  12. package/docs/migration.md +75 -0
  13. package/docs/quality-gates.pt-BR.md +160 -0
  14. package/docs/quality.md +56 -0
  15. package/docs/releases.pt-BR.md +186 -0
  16. package/docs/setup-documents.pt-BR.md +120 -0
  17. package/docs/usage.pt-BR.md +305 -0
  18. package/docs/validation.md +95 -0
  19. package/lib/adapters.js +115 -0
  20. package/lib/check.js +221 -0
  21. package/lib/context.js +117 -0
  22. package/lib/documents.js +71 -0
  23. package/lib/handoff.js +81 -0
  24. package/lib/history.js +16 -0
  25. package/lib/install.js +181 -0
  26. package/lib/io.js +119 -0
  27. package/lib/legacy-hashes.json +31 -0
  28. package/lib/markdown.js +71 -0
  29. package/lib/project.js +200 -0
  30. package/lib/quality.js +24 -0
  31. package/lib/revisions.js +79 -0
  32. package/lib/verification.js +267 -0
  33. package/package.json +55 -32
  34. package/templates/AGENTS.md +12 -50
  35. package/templates/QUICKSTART.md +16 -70
  36. package/templates/agents/AGENTS.md +54 -92
  37. package/templates/agents/DECISIONS.md +5 -21
  38. package/templates/agents/POLICIES.md +14 -0
  39. package/templates/agents/PROJECT.md +35 -145
  40. package/templates/agents/QUALITY.md +19 -0
  41. package/templates/agents/RULES.md +22 -132
  42. package/templates/agents/SETUP.md +38 -224
  43. package/templates/agents/config.yaml +13 -0
  44. package/templates/agents/current/_template.md +19 -0
  45. package/templates/agents/decisions/_template.md +22 -0
  46. package/templates/agents/prompts/conventional-commit.md +11 -51
  47. package/templates/agents/prompts/intake.md +23 -0
  48. package/templates/agents/prompts/pr-template.md +12 -21
  49. package/templates/agents/prompts/quality.md +57 -0
  50. package/templates/agents/prompts/rpi-close.md +23 -0
  51. package/templates/agents/prompts/rpi-implement.md +31 -68
  52. package/templates/agents/prompts/rpi-plan.md +87 -117
  53. package/templates/agents/prompts/rpi-research.md +21 -67
  54. package/templates/agents/prompts/rpi-validate.md +40 -103
  55. package/templates/agents/prompts/skill-call.md +11 -29
  56. package/templates/agents/prompts/task-create.md +13 -42
  57. package/templates/agents/skills/_template/SKILL.md +11 -37
  58. package/templates/agents/harness/claude-code/.hooks/guard-delete.sh +0 -22
  59. package/templates/agents/harness/claude-code/.hooks/guard-deps.sh +0 -25
  60. package/templates/agents/harness/claude-code/.hooks/guard-git.sh +0 -24
  61. package/templates/agents/harness/claude-code/commands/sdd-implement.md +0 -3
  62. package/templates/agents/harness/claude-code/commands/sdd-plan.md +0 -3
  63. package/templates/agents/harness/claude-code/commands/sdd-research.md +0 -3
  64. package/templates/agents/harness/claude-code/commands/sdd-validate.md +0 -3
  65. package/templates/agents/harness/claude-code/settings.json +0 -34
  66. package/templates/agents/harness/opencode/commands/sdd-implement.md +0 -3
  67. package/templates/agents/harness/opencode/commands/sdd-plan.md +0 -3
  68. package/templates/agents/harness/opencode/commands/sdd-research.md +0 -3
  69. package/templates/agents/harness/opencode/commands/sdd-validate.md +0 -3
  70. package/templates/agents/harness/opencode/opencode.json +0 -18
@@ -1,117 +1,87 @@
1
- # Prompt: Plan (RPI — Phase 2)
2
-
3
- Use this prompt to generate all planning artifacts for a spec.
4
-
5
- ---
6
-
7
- ## Agent instruction
8
-
9
- You will plan the implementation of an already-researched feature.
10
-
11
- **Before starting, read:**
12
- - `agents/AGENTS.md`
13
- - `agents/PROJECT.md`
14
- - `agents/RULES.md`
15
- - `agents/specs/NNN-name/RESEARCH.md`
16
- - Check which files already exist in `agents/specs/NNN-name/`
17
-
18
- **Expected input:**
19
- - Spec number and name (e.g. `001-user-authentication`)
20
-
21
- ---
22
-
23
- ## Adaptive behavior
24
-
25
- | Situation | Action |
26
- |---|---|
27
- | SPEC.md does not exist | Create all artifacts from scratch |
28
- | SPEC.md exists, TASK.md does not | Generate only TASK.md and PROGRESS.md based on existing SPEC.md |
29
- | All artifacts exist | Review and update only what is incomplete or outdated |
30
-
31
- ---
32
-
33
- ## Deliverables
34
-
35
- Generate the following files inside `agents/specs/NNN-name/`:
36
-
37
- ### SPEC.md
38
- ```markdown
39
- # SPEC — [Feature Name]
40
-
41
- ## Goal
42
- What this feature delivers (refined from RESEARCH.md).
43
-
44
- ## Entities involved
45
- Data structures created or modified, with full field list.
46
-
47
- ## Business Rules
48
- Numbered list. Only rules that directly affect implementation.
49
-
50
- ## Expected behaviors
51
- Per behavior: what triggers it, what it does, what the output is.
52
-
53
- ## UI
54
- Describe components, states, interactions and validations.
55
- Write `N/A` if no UI is involved.
56
-
57
- ## Out of scope
58
- Explicit list of what this spec does NOT cover.
59
- ```
60
-
61
- ### TASK.md
62
- ```markdown
63
- # TASK — [Feature Name]
64
-
65
- ## Tasks
66
-
67
- - [ ] T01 — [Verb + object + target file] (e.g. Create User entity schema in config)
68
- verify: [observable condition confirming completion — e.g. endpoint returns 201 with `id` field; returns 400 if `email` is missing]
69
- - [ ] T02 — ...
70
- verify: [specific and testable condition — never "works correctly"]
71
- ```
72
-
73
- Task rules:
74
- - One task = one file created/modified, or one isolated function
75
- - Order must respect dependencies (schema → data layer → service → route → UI)
76
- - No circular dependencies between tasks
77
- - Preferred verbs: `Add`, `Create`, `Implement`, `Update`, `Extract`, `Register`
78
- - Each task must include a `verify:` line with a specific, observable condition — never "works correctly"
79
-
80
- ### PROGRESS.md
81
- ```markdown
82
- # PROGRESS — [Feature Name]
83
-
84
- ## Status: 🔴 Not started
85
-
86
- ## Tasks
87
-
88
- | Task | Status | Notes |
89
- |---|---|---|
90
- | T01 | Pending | |
91
- | T02 | Pending | |
92
-
93
- ## History
94
-
95
- (updated during implementation)
96
- ```
97
-
98
- ### TEST.md
99
- ```markdown
100
- # TEST — [Feature Name]
101
-
102
- ## Acceptance Test Cases
103
-
104
- ### TC01 — [Behavior being tested]
105
- - **Given** [initial state]
106
- - **When** [action performed]
107
- - **Then** [expected outcome]
108
- ```
109
-
110
- ---
111
-
112
- ## Constraints
113
-
114
- - Do not write code in this phase
115
- - Do not implement any task — only plan
116
- - If the RESEARCH.md has unresolved open questions, **stop and ask** before planning
117
- - After delivering all artifacts, **wait for approval** before advancing to Implement
1
+ ---
2
+ schema_version: 2
3
+ kind: procedure
4
+ phase: plan
5
+ description: Define scoped requirements, acceptance and cohesive tasks.
6
+ ---
7
+
8
+ # Plan
9
+
10
+ Read the actual sources, applicable current capability contracts and code. Run sdd context ID --phase plan when available.
11
+ Use the change mode from change.yaml. Resolve only ambiguities that affect behavior or scope; state routine technical assumptions.
12
+ For a multi-feature request or oversized context packet, follow intake.md and propose cohesive linked deliveries.
13
+
14
+ ## Contract
15
+
16
+ SPEC.md front matter is the structured contract. Keep rationale and prose detail in its body.
17
+ Preserve IDs when editing. Use local IDs qualified externally by the change ID.
18
+
19
+ ~~~yaml
20
+ ---
21
+ schema_version: 2
22
+ artifact: spec
23
+ id: 001-example
24
+ requirements:
25
+ - id: REQ01
26
+ description: State an observable requirement, including important failure behavior.
27
+ acceptance:
28
+ - id: AC01
29
+ covers: [REQ01]
30
+ description: State concrete inputs and expected outcomes.
31
+ method: automated
32
+ checks: [unit]
33
+ ---
34
+ ~~~
35
+
36
+ Automated checks must name real checks configured in agents/config.yaml.
37
+ Use method: manual or inspection for cases that require observed evidence rather than a command.
38
+ Automated command success is execution evidence; review assertions to establish semantic coverage.
39
+
40
+ ## Tasks
41
+
42
+ Standard and rigorous modes use TASK.md. Light mode may omit it.
43
+ Keep task status only in its front matter. One task is a cohesive verifiable outcome and may change multiple files.
44
+
45
+ ~~~yaml
46
+ ---
47
+ schema_version: 2
48
+ artifact: tasks
49
+ id: 001-example
50
+ tasks:
51
+ - id: T01
52
+ title: Implement the behavior and its regression checks.
53
+ status: pending
54
+ covers: [REQ01, AC01]
55
+ depends_on: []
56
+ verify: The acceptance scenario and relevant regression checks pass.
57
+ ---
58
+ ~~~
59
+
60
+ Plan tests with implementation; permit TDD and incremental verification.
61
+ Use concrete initial-state/action/outcome examples in acceptance, including relevant failure/boundary cases.
62
+ Follow config quality.tdd; use quality.md when defining strategy or project-specific gates.
63
+ Classify external checks/criteria accurately before approval. Separate implementation from external-only verification tasks.
64
+ Missing environment access calls for a safe user handoff, not mandatory production testing or an impossible completion gate.
65
+ Allow many-to-many relationships among requirements, acceptance, tasks and tests.
66
+ Retain cancelled records with status: cancelled and cancellation containing reason and by.
67
+ Do not infer that cancellation requests deleting code.
68
+
69
+ ## Context and authorization
70
+
71
+ In change.yaml select capabilities, decisions, skills and dependency spec IDs.
72
+ Use context entries as a project-relative path or {path: ..., section: Exact heading}.
73
+ Keep sources as paths or explicit path/section references to original requirements, which remain discoverable.
74
+ Do not populate every project artifact merely because a template exists.
75
+ For meaningful changes, add a short impact/regression note in SPEC's body: affected consumers/shared contracts,
76
+ behavior to preserve, relevant regression checks and any migration/cache/integration considerations.
77
+ Inspect only applicable areas; a small isolated change may need one sentence. Do not create a duplicate impact document.
78
+
79
+ Run sdd check ID and fix structural gaps. Explain the outcome, assumptions and material choices.
80
+ When the user has authorized this scope, record it with sdd approve ID --by NAME.
81
+ Never claim a human authorization that did not occur. Follow the existing conversation's authorization;
82
+ do not repeatedly ask for permission already given for the same scope.
83
+
84
+ When an approved business rule or policy changes, update the contract and affected task definitions while
85
+ preserving IDs. Run sdd revise ID --by NAME --reason REASON --dry-run to inspect impact, then record the revision.
86
+ Affected completed tasks and their dependents are reopened; policy/prose changes may conservatively reopen all.
87
+ Review and authorize the revised scope before continuing. Use a new linked spec for a separate deliverable.
@@ -1,67 +1,21 @@
1
- # Prompt: Research (RPI — Phase 1)
2
-
3
- Use this prompt to analyze the scope of a feature and produce RESEARCH.md.
4
-
5
- ---
6
-
7
- ## Agent instruction
8
-
9
- You will analyze the scope of a feature and produce the `RESEARCH.md` for the spec.
10
- This file is the only artifact of this phase and will be the context for the Plan phase.
11
-
12
- **Before starting, read:**
13
- - `agents/AGENTS.md`
14
- - `agents/PROJECT.md`
15
- - `agents/RULES.md`
16
-
17
- **Expected input:**
18
- - Requirements source (PRD, textual description, or any document provided)
19
- - Spec name and number (e.g. `001-user-authentication`)
20
-
21
- ---
22
-
23
- ## Deliverable: RESEARCH.md
24
-
25
- Produce **only** the file `agents/specs/NNN-name/RESEARCH.md` with the structure below.
26
- Be concise — this file will be read in future sessions as compressed context.
27
-
28
- ```markdown
29
- # RESEARCH — [Feature Name]
30
-
31
- ## Goal
32
- One clear sentence stating what this feature delivers.
33
-
34
- ## Entities and data structures
35
- List the entities involved, data structures to be created or modified,
36
- and their expected fields/schema.
37
-
38
- ## Relevant business rules
39
- Only the rules that directly impact implementation.
40
- Reference by number if the PRD uses numbered rules.
41
-
42
- ## Dependencies
43
- Features or specs that must be complete before this one.
44
- List `None` if there are none.
45
-
46
- ## Scope decisions
47
- What is explicitly inside and outside the scope of this spec.
48
- Be explicit about edge cases.
49
-
50
- ## Risks and attention points
51
- Resolved ambiguities, identified trade-offs, and warnings for the implementation phase.
52
-
53
- ## Open questions
54
- List only questions that block the Plan phase.
55
- If none, write `None`.
56
- ```
57
-
58
- ---
59
-
60
- ## Constraints
61
-
62
- - Produce only RESEARCH.md — no other file must be created
63
- - Do not create SPEC.md, TASK.md, PROGRESS.md, TEST.md or any code file
64
- - Do not write code in this phase
65
- - Do not anticipate implementation decisions
66
- - If there are open questions, **stop and wait for answers** before delivering RESEARCH.md
67
- - After delivering RESEARCH.md, **wait for approval** before advancing to Plan
1
+ ---
2
+ schema_version: 2
3
+ kind: procedure
4
+ phase: research
5
+ description: Resolve uncertainty using requirements sources and the actual code.
6
+ ---
7
+
8
+ # Research
9
+
10
+ Use the active change manifest and project map. Inspect only the modules, source requirements and integrations relevant to this change.
11
+ Preserve source paths and distinguish facts, assumptions, alternatives and unresolved questions.
12
+ Do not replace the source requirements with an unsupported summary.
13
+ For a change to existing behavior, inspect direct consumers, shared contracts and relevant callers/tests,
14
+ not only the function being edited. Establish available baseline results; separate pre-existing failures,
15
+ introduced regressions and unavailable-environment checks. Keep the impact note short and source-linked.
16
+
17
+ Create RESEARCH.md only when research adds useful information. It is required in rigorous mode.
18
+ Use front matter with schema_version: 2, artifact: research and id matching the change ID.
19
+ In the body record the goal, relevant evidence, scope, risks and unresolved blocking questions.
20
+ Do not implement code. If a material question blocks planning, ask it while continuing independent investigation.
21
+ No separate approval of research/progress/empty decisions files is required; authorization applies to the normative scope bundle.
@@ -1,103 +1,40 @@
1
- # Prompt: Validate (RPI — Phase 4)
2
-
3
- Use this prompt to verify that the implementation satisfies the spec and all acceptance criteria.
4
-
5
- ---
6
-
7
- ## Agent instruction
8
-
9
- You will validate the implementation of a completed feature against its spec and test cases.
10
-
11
- **Before starting, read:**
12
- - `agents/AGENTS.md`
13
- - `agents/PROJECT.md`
14
- - `agents/RULES.md`
15
- - `agents/specs/NNN-name/SPEC.md`
16
- - `agents/specs/NNN-name/TASK.md` — identify which tasks are complete `[x]` and which are pending `[ ]`
17
- - `agents/specs/NNN-name/TEST.md`
18
- - `agents/specs/NNN-name/PROGRESS.md`
19
- - The code files produced during Implement (listed in PROGRESS.md History)
20
-
21
- **Expected input:**
22
- - Spec number and name (e.g. `001-user-authentication`)
23
-
24
- ---
25
-
26
- ## Validation procedure
27
-
28
- ### Step 1 — Determine validation scope
29
- Validation may be requested at any time — it does not require all tasks to be complete.
30
-
31
- - If all tasks are `[x]`: full validation — cover every task and every test case in TEST.md
32
- - If some tasks are still `[ ]`: partial validation — scope to completed tasks only; explicitly list which tasks and test cases are **out of scope** in VALIDATION.md
33
-
34
- Never block validation because tasks are pending. Report the scope clearly instead.
35
-
36
- ### Step 2 — Verify task-level conditions
37
- For each task in `TASK.md`, read its `verify:` condition and check whether
38
- the produced code satisfies it. Examine the actual files — do not assume.
39
-
40
- ### Step 3 — Verify acceptance test cases
41
- For each test case in `TEST.md`, assess whether the implementation satisfies
42
- the Given/When/Then scenario. Read the relevant code directly.
43
-
44
- ### Step 4 — Produce VALIDATION.md
45
- Write `agents/specs/NNN-name/VALIDATION.md` with the structure below.
46
-
47
- ---
48
-
49
- ## Deliverable: VALIDATION.md
50
-
51
- ```markdown
52
- # VALIDATION — [Feature Name]
53
-
54
- ## Status: [✅ Approved | ⚠️ Partial | ❌ Failed]
55
-
56
- ## Scope
57
- [Full — all tasks complete] or [Partial — tasks T01..TNN validated; TXX..TYY pending implementation]
58
-
59
- ---
60
-
61
- ## Task Verification
62
-
63
- | Task | verify: condition | Status |
64
- |---|---|---|
65
- | T01 | [condition from TASK.md] | ✅ Met / ❌ Not met / ⚠️ Partial |
66
- | T02 | ... | ... |
67
-
68
- ---
69
-
70
- ## Acceptance Test Coverage
71
-
72
- | Test Case | Description | Status | Evidence |
73
- |---|---|---|---|
74
- | TC01 | [behavior being tested] | ✅ Verified / ❌ Not met / ⚠️ Partial | [file and line or behavior observed in code] |
75
- | TC02 | ... | ... | ... |
76
-
77
- ---
78
-
79
- ## Gaps Found
80
-
81
- List any `verify:` conditions or test cases that are not fully met.
82
- Write `None` if everything is satisfied.
83
-
84
- - [ ] [Gap description — what is missing or incorrect]
85
-
86
- ---
87
-
88
- ## Recommendation
89
-
90
- [Approve — all criteria met]
91
- [Request fixes — list the gaps that must be resolved before approval]
92
- ```
93
-
94
- ---
95
-
96
- ## Constraints
97
-
98
- - Do not write or modify any code in this phase
99
- - Do not mark gaps as acceptable — report them, even if minor
100
- - If a test case cannot be verified from static code analysis alone, flag it explicitly:
101
- `⚠️ Requires runtime verification — cannot confirm from code alone`
102
- - After delivering VALIDATION.md, **wait for developer review**
103
- - If gaps are found and fixed, a new Validate session must be run — do not reuse this report
1
+ ---
2
+ schema_version: 2
3
+ kind: procedure
4
+ phase: validate
5
+ description: Verify actual behavior and record execution tied to the tested inputs.
6
+ ---
7
+
8
+ # Validate
9
+
10
+ Choose partial feedback or final verification explicitly. Pending tasks do not block partial verification.
11
+ Review the requirement-to-acceptance links and actual test assertions, failure cases and relevant integrations.
12
+ Compare available baseline results with the changed behavior and affected consumers. Identify pre-existing issues
13
+ without silently weakening required acceptance, changing expected results, or expanding scope into unrelated repairs.
14
+ Inspect the diff, existing work and untracked files; do not attribute unrelated changes to this spec without evidence.
15
+
16
+ Run sdd verify ID --scope partial for diagnostic feedback or sdd verify ID for the final check.
17
+ The CLI executes only the configured commands. It records exit codes, timeouts, logs and an input digest.
18
+ A failed required command fails verification. Advisory failures remain visible warnings.
19
+ External commands are skipped unless --run-external is explicitly authorized. An unexecuted case remains unverified.
20
+
21
+ For manual/inspection cases, run sdd evidence ID to create a template for current inputs.
22
+ Execute the scenario or inspect the relevant code, then record result, by, at and notes in EVIDENCE.json.
23
+ Reference supporting artifacts when available. Run sdd verify ID --evidence PATH.
24
+ A manual case can pass; its method and result are separate fields.
25
+ If inputs changed, regenerate evidence and repeat affected checks rather than reusing an old result.
26
+ The evidence command preserves existing files and returns a new path when necessary; use that path with --evidence.
27
+
28
+ During validation, preserve the code and contract under review. Report fixes for implementation and rerun verification after fixes.
29
+ Tests may produce expected artifacts in an isolated environment; they are not intrinsically read-only.
30
+ Do not run production operations or install dependencies merely to satisfy validation.
31
+
32
+ Separate attribution, severity and impact on the intended delivery. Keep scope controlled while escalating findings
33
+ that compromise the delivery, even when their origin predates this change.
34
+ A passing receipt records execution and declared links; it does not prove that every assertion matches the intended behavior.
35
+
36
+ For unavailable environments follow quality.md and sdd handoff ID. Fill the generated JSON and run
37
+ sdd verify ID --handoff PATH. USER_TESTS.md contains the recorded steps and safety constraints.
38
+ A full limited result permits implementation closure while retaining pending verification; it never means passed external tests.
39
+ If no safe test is possible without production side effects, record unavailable and the condition for resuming.
40
+ Known failures for the same inputs remain failures when evidence or external execution is omitted later.
@@ -1,29 +1,11 @@
1
- # Prompt: Execute task using a Skill
2
-
3
- Use this prompt when a task requires following a specific skill pattern.
4
-
5
- ---
6
-
7
- ## Agent instruction
8
-
9
- Read:
10
- - `agents/AGENTS.md`
11
- - `agents/PROJECT.md`
12
- - `agents/RULES.md`
13
- - `agents/specs/<SPEC_NUMBER>/SPEC.md`
14
- - `agents/specs/<SPEC_NUMBER>/TASK.md`
15
- - `agents/specs/<SPEC_NUMBER>/PROGRESS.md`
16
- - `agents/skills/<SKILL_NAME>/SKILL.md`
17
-
18
- Now implement ONLY the task: `<TASK_ID>`
19
-
20
- Use the skill as the mandatory pattern source for this task.
21
-
22
- ---
23
-
24
- ## Rules
25
-
26
- - Do not run git commands
27
- - Do not add comments or documentation unless explicitly required by the skill
28
- - Update PROGRESS.md upon completion
29
- - Stop and wait for confirmation before the next task
1
+ ---
2
+ schema_version: 2
3
+ kind: utility
4
+ description: Apply a selected project skill during the canonical implementation procedure.
5
+ ---
6
+
7
+ # Apply a skill
8
+
9
+ Use sdd skills list to discover metadata, then sdd skills show NAME for the selected body.
10
+ Follow rpi-implement.md for the requested spec/task. A skill does not bypass scope approval,
11
+ execution permissions or verification. Read referenced resources only when they are relevant.
@@ -1,42 +1,13 @@
1
- # Prompt: Task Create
2
-
3
- Use this prompt to create or revise the TASK.md of an existing spec.
4
-
5
- ---
6
-
7
- ## Agent instruction
8
-
9
- From an already approved SPEC.md, generate or update the TASK.md with a proper task breakdown.
10
-
11
- **Expected input:**
12
- - Content or path of the SPEC.md
13
-
14
- ---
15
-
16
- ## Task generation rules
17
-
18
- - One task = one file created or modified, or one isolated function
19
- - Order must respect dependencies:
20
- - Schema / config → data access → business logic → routing → UI
21
- - Tasks that are consumed by other tasks must come first
22
- - No task should have a circular dependency on another
23
- - Each task must be independently reviewable and verifiable
24
-
25
- ## Required format
26
-
27
- ```
28
- # TASK — [Feature Name]
29
-
30
- ## Tasks
31
-
32
- - [ ] T01 — [Verb + object + target file] (e.g. Add User schema to config module)
33
- - [ ] T02 — ...
34
- ```
35
-
36
- ## Preferred verbs
37
-
38
- `Add`, `Create`, `Implement`, `Update`, `Extract`, `Register`, `Remove`, `Move`
39
-
40
- ---
41
-
42
- **After generating TASK.md, stop and wait for approval before implementing.**
1
+ ---
2
+ schema_version: 2
3
+ kind: utility
4
+ description: Revise the canonical task plan without maintaining a parallel format.
5
+ ---
6
+
7
+ # Revise tasks
8
+
9
+ Follow rpi-plan.md and the existing SPEC.md contract.
10
+ Preserve task IDs. Use canonical TASK.md front matter with title, status, covers, depends_on and verify.
11
+ Add cohesive verifiable tasks rather than a task per file. Do not duplicate status in PROGRESS.md.
12
+ Changes to the normative task plan invalidate scope approval. Run sdd check, explain the change
13
+ and record actual user authorization for the revised scope before implementation when required.
@@ -1,37 +1,11 @@
1
- # Skill: [Capability Name]
2
-
3
- ## When to use this skill
4
-
5
- [Specific task types that trigger this skill. Be precise — the agent uses this section
6
- to decide whether to load the skill. Vague descriptions reduce detection accuracy.]
7
-
8
- ---
9
-
10
- ## Patterns
11
-
12
- [Correct approach with code examples from this project's stack]
13
-
14
- ```language
15
- // example
16
- ```
17
-
18
- ---
19
-
20
- ## Anti-patterns
21
-
22
- | Anti-pattern | Why to avoid |
23
- |---|---|
24
- | [What NOT to do] | [Reason] |
25
-
26
- ---
27
-
28
- ## Constraints
29
-
30
- - [Hard rule specific to this capability]
31
-
32
- ---
33
-
34
- ## Checklist
35
-
36
- - [ ] [Item the agent must verify before finishing a task using this skill]
37
- - [ ] [Another verification item]
1
+ ---
2
+ name: project-capability
3
+ description: Describe the recurring procedure and the specific task types that should activate it.
4
+ ---
5
+
6
+ # Project capability
7
+
8
+ Replace the name and rename the containing folder to match it.
9
+ State the necessary procedure, project-specific constraints and observable completion checks.
10
+ Link longer examples, scripts or domain references relative to this directory.
11
+ Keep discovery conditions in description, not in an arbitrary body excerpt.
@@ -1,22 +0,0 @@
1
- #!/bin/bash
2
- # guard-delete.sh
3
- # Asks for confirmation before running destructive delete commands (rm -rf, rm -r, rm -f).
4
- # Exit 0 = allow | Exit 2 = block (Claude Code convention)
5
-
6
- INPUT=$(cat)
7
-
8
- COMMAND=$(echo "$INPUT" | grep -o '"command":"[^"]*"' | head -1 | sed 's/"command":"//;s/"//')
9
-
10
- # Intercept rm with recursive or force flags
11
- if echo "$COMMAND" | grep -qE "rm\s+(-[a-zA-Z]*[rf][a-zA-Z]*|--recursive|--force)"; then
12
- echo "⚠️ Agent wants to run: $COMMAND" >&2
13
- read -r -p "Allow destructive delete? [y/N] " response < /dev/tty
14
- if [[ "$response" =~ ^[Yy]$ ]]; then
15
- exit 0
16
- else
17
- echo "Blocked by guard-delete.sh" >&2
18
- exit 2
19
- fi
20
- fi
21
-
22
- exit 0
@@ -1,25 +0,0 @@
1
- #!/bin/bash
2
- # guard-deps.sh
3
- # Asks for confirmation before installing new packages.
4
- # Reinforces R08: no new dependency without justification in DECISIONS.md.
5
- # Exit 0 = allow | Exit 2 = block (Claude Code convention)
6
-
7
- INPUT=$(cat)
8
-
9
- COMMAND=$(echo "$INPUT" | grep -o '"command":"[^"]*"' | head -1 | sed 's/"command":"//;s/"//')
10
-
11
- # Intercept install commands that add NEW packages (not bare installs like "npm install" with no args)
12
- if echo "$COMMAND" | grep -qE "^(npm install|npm i|yarn add|pnpm add|bun add) .+" || \
13
- echo "$COMMAND" | grep -qE "^(pip install|pip3 install|uv add) .+"; then
14
- echo "⚠️ Agent wants to install a new dependency: $COMMAND" >&2
15
- echo " Remember: R08 requires justification in DECISIONS.md before adding packages." >&2
16
- read -r -p "Allow? [y/N] " response < /dev/tty
17
- if [[ "$response" =~ ^[Yy]$ ]]; then
18
- exit 0
19
- else
20
- echo "Blocked by guard-deps.sh" >&2
21
- exit 2
22
- fi
23
- fi
24
-
25
- exit 0