@bcelep/capint 0.4.2
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.
- package/AGENT.md +28 -0
- package/CHANGELOG.md +58 -0
- package/README.md +94 -0
- package/bin/capint.js +90 -0
- package/design.md +23 -0
- package/docs/architecture-decisions.md +95 -0
- package/docs/execution-intent-contract.md +81 -0
- package/docs/manifest-schema.md +36 -0
- package/docs/release-checklist.md +31 -0
- package/package.json +33 -0
- package/projections/session-start.md +32 -0
- package/registry.json +12 -0
- package/scripts/release-check.mjs +40 -0
- package/scripts/validate-matrix.mjs +83 -0
- package/skill-routing-matrix.json +150 -0
- package/skills/capability-router/SKILL.md +3 -0
- package/skills/context-memory-bridge/SKILL.md +17 -0
- package/skills/localization-hub/SKILL.md +17 -0
- package/skills/refactor/SKILL.md +17 -0
- package/skills/systematic-debugging/SKILL.md +19 -0
- package/src/commands/audit.js +32 -0
- package/src/commands/consult.js +64 -0
- package/src/commands/doctor.js +36 -0
- package/src/commands/ide.js +62 -0
- package/src/commands/init.js +50 -0
- package/src/commands/memory.js +60 -0
- package/src/commands/route.js +30 -0
- package/src/commands/scaffold.js +37 -0
- package/src/commands/status.js +22 -0
- package/src/commands/uninstall.js +46 -0
- package/src/commands/upgrade.js +69 -0
- package/src/lib/audit.js +107 -0
- package/src/lib/capability-router.js +118 -0
- package/src/lib/context-memory-bridge.js +87 -0
- package/src/lib/context-pack.js +39 -0
- package/src/lib/contract.js +71 -0
- package/src/lib/doctor.js +115 -0
- package/src/lib/event-log.js +40 -0
- package/src/lib/execution-policy.js +168 -0
- package/src/lib/ide-sync.js +277 -0
- package/src/lib/intent-parser.js +116 -0
- package/src/lib/orchestration.js +25 -0
- package/src/lib/providers/activation-policy.js +93 -0
- package/src/lib/providers/graph-provider.js +35 -0
- package/src/lib/providers/local-graph-adapter.js +40 -0
- package/src/lib/providers/local-memory-adapter.js +41 -0
- package/src/lib/providers/memory-provider.js +35 -0
- package/src/lib/route-engine.js +191 -0
- package/src/lib/scaffold/file-policy.js +92 -0
- package/src/lib/scaffold/index.js +29 -0
- package/src/lib/scaffold/manifest-schema.js +116 -0
- package/src/lib/scaffold/manifest.js +34 -0
- package/src/lib/scaffold/presets.js +132 -0
- package/src/lib/uninstall.js +126 -0
- package/src/lib/upgrade-matrix.js +120 -0
- package/templates/bundle/skills/capability-router/SKILL.md +12 -0
- package/templates/bundle/skills/context-memory-bridge/SKILL.md +17 -0
- package/templates/bundle/skills/localization-hub/SKILL.md +17 -0
- package/templates/bundle/skills/refactor/SKILL.md +17 -0
- package/templates/bundle/skills/systematic-debugging/SKILL.md +19 -0
- package/templates/bundle/workflows/forge.md +51 -0
- package/templates/minimal/.capint/rules/core.md +18 -0
- package/templates/minimal/AGENT.md +26 -0
- package/templates/minimal/AGENTS.md +9 -0
- package/templates/minimal/design.md +25 -0
- package/templates/minimal/registry.json +7 -0
- package/templates/prismx-compatible/.capint/context.json +8 -0
- package/templates/prismx-compatible/HANDOFF.md +19 -0
- package/workflows/forge.md +51 -0
|
@@ -0,0 +1,18 @@
|
|
|
1
|
+
# CapInt core rules
|
|
2
|
+
|
|
3
|
+
<!-- capint:managed:start -->
|
|
4
|
+
## Session
|
|
5
|
+
|
|
6
|
+
1. Read `AGENT.md` and `design.md` before non-trivial edits.
|
|
7
|
+
2. Emit Execution Intent + Confirm for medium/heavy tasks.
|
|
8
|
+
|
|
9
|
+
## Safety
|
|
10
|
+
|
|
11
|
+
- No blind overwrite of existing docs.
|
|
12
|
+
- No claiming done without verification evidence on medium/heavy tasks.
|
|
13
|
+
- Memory/graph providers default off unless policy activates them.
|
|
14
|
+
|
|
15
|
+
## Routing
|
|
16
|
+
|
|
17
|
+
- Capability-first; slash commands are override only.
|
|
18
|
+
<!-- capint:managed:end -->
|
|
@@ -0,0 +1,26 @@
|
|
|
1
|
+
# AGENT.md — {{project_name}}
|
|
2
|
+
|
|
3
|
+
CapInt capability-first orchestration contract for this project.
|
|
4
|
+
|
|
5
|
+
<!-- capint:managed:start -->
|
|
6
|
+
## Task contract
|
|
7
|
+
|
|
8
|
+
For every task:
|
|
9
|
+
1. Build Execution Intent
|
|
10
|
+
2. Ask confirm options (`apply_now | plan_first | analyze_only`)
|
|
11
|
+
3. Route via capability + provider resolution
|
|
12
|
+
|
|
13
|
+
## Runtime chain
|
|
14
|
+
|
|
15
|
+
`parseIntent -> routeCapability -> resolveMemoryStrategy -> buildExecutionPolicy -> formatOutput`
|
|
16
|
+
|
|
17
|
+
## Scope
|
|
18
|
+
|
|
19
|
+
- Chat-first UX; no skill name memorization.
|
|
20
|
+
- Memory strategy is silent (`none|optional|required`).
|
|
21
|
+
- Verification required on medium/heavy before claiming done.
|
|
22
|
+
<!-- capint:managed:end -->
|
|
23
|
+
|
|
24
|
+
## Project notes
|
|
25
|
+
|
|
26
|
+
Add project-specific agent guidance below this line (never overwritten by scaffold).
|
|
@@ -0,0 +1,9 @@
|
|
|
1
|
+
<!-- capint:managed:start -->
|
|
2
|
+
<!-- source: capint init | AGENTS.md stub -->
|
|
3
|
+
# AGENTS.md — {{project_name}}
|
|
4
|
+
|
|
5
|
+
Entry point for Codex / Cursor agents. Session protocol: **AGENT.md**
|
|
6
|
+
|
|
7
|
+
- Router: `capint route "<task>"`
|
|
8
|
+
- Health: `capint doctor`
|
|
9
|
+
<!-- capint:managed:end -->
|
|
@@ -0,0 +1,25 @@
|
|
|
1
|
+
# design.md — {{project_name}}
|
|
2
|
+
|
|
3
|
+
<!-- capint:managed:start -->
|
|
4
|
+
## Goal
|
|
5
|
+
|
|
6
|
+
Describe the primary outcome this project must deliver.
|
|
7
|
+
|
|
8
|
+
## Boundaries
|
|
9
|
+
|
|
10
|
+
- In scope:
|
|
11
|
+
- Out of scope:
|
|
12
|
+
|
|
13
|
+
## Success criteria
|
|
14
|
+
|
|
15
|
+
- [ ] Measurable outcome 1
|
|
16
|
+
- [ ] Measurable outcome 2
|
|
17
|
+
|
|
18
|
+
## Updated
|
|
19
|
+
|
|
20
|
+
{{date}}
|
|
21
|
+
<!-- capint:managed:end -->
|
|
22
|
+
|
|
23
|
+
## Notes
|
|
24
|
+
|
|
25
|
+
Project-specific design decisions (safe from scaffold overwrite unless managed block updated).
|
|
@@ -0,0 +1,51 @@
|
|
|
1
|
+
# /forge
|
|
2
|
+
|
|
3
|
+
Default implementation workflow for medium/heavy tasks after confirm.
|
|
4
|
+
|
|
5
|
+
Inspired by harness-style phased orchestration; CapInt router selects this workflow for debug/refactor capabilities.
|
|
6
|
+
|
|
7
|
+
## Execution modes (from confirm option)
|
|
8
|
+
|
|
9
|
+
| Confirm option | Mode | Behavior |
|
|
10
|
+
|----------------|------|----------|
|
|
11
|
+
| `apply_now` | subagent | Single focused pass, minimal ceremony |
|
|
12
|
+
| `plan_first` | pipeline | Phased plan before edits |
|
|
13
|
+
| `analyze_only` | explore | Read-only analysis, no file writes |
|
|
14
|
+
|
|
15
|
+
## Phase 0 — Context
|
|
16
|
+
|
|
17
|
+
1. Read `AGENT.md`, `design.md`, optional `HANDOFF.md`
|
|
18
|
+
2. Use `context_pack` from Execution Intent when memory is optional/required
|
|
19
|
+
3. If ambiguous, stop and ask one clarifying question
|
|
20
|
+
|
|
21
|
+
## Phase 1 — Intent lock
|
|
22
|
+
|
|
23
|
+
1. Restate capability + resolution from Execution Intent
|
|
24
|
+
2. List assumptions and success criteria
|
|
25
|
+
3. Wait for confirm unless `apply_now` on light tasks
|
|
26
|
+
|
|
27
|
+
## Phase 2 — Execute
|
|
28
|
+
|
|
29
|
+
1. Apply smallest correct change set
|
|
30
|
+
2. Match project conventions; no drive-by refactors
|
|
31
|
+
3. For `producer_reviewer` pattern: implement then self-review before claiming done
|
|
32
|
+
|
|
33
|
+
## Phase 3 — Verify
|
|
34
|
+
|
|
35
|
+
1. Run verification hints from Execution Intent
|
|
36
|
+
2. Evidence before claims — no "done" without command output
|
|
37
|
+
3. If verification fails, return to Phase 2 (max 2 retries)
|
|
38
|
+
|
|
39
|
+
## Phase 4 — Handoff
|
|
40
|
+
|
|
41
|
+
1. Summarize what changed and why
|
|
42
|
+
2. Note follow-on capabilities from `orchestration.chains` if any
|
|
43
|
+
3. Update `HANDOFF.md` when session material decisions were made
|
|
44
|
+
|
|
45
|
+
## Error handling
|
|
46
|
+
|
|
47
|
+
| Situation | Action |
|
|
48
|
+
|-----------|--------|
|
|
49
|
+
| Blocked on missing info | `NEEDS_CONTEXT` — ask user, do not guess |
|
|
50
|
+
| Verification failed | Fix and re-verify; do not skip |
|
|
51
|
+
| Scope creep detected | Split task; route again via `capint consult` |
|