Skip to content

stevesolun/ctx / docs

skill-router

Flagged

Repo-aware recommendation manager for ctx. Scans the active repository, identifies stack and workflow signals, recommends a capped set of skills, agents, and MCP servers, and unloads helpers that no longer match the current work after user confirmation. Harnesses are recommended by the custom-model onboarding flow or loop adapters and then attach to the same recommendation layer.

FileKindSize
SKILL.mdskill md9.3 KB
assets/javascripts/catalog.jsscript2.5 KB
assets/javascripts/repo-stats-refresh.jsscript3.2 KB
backup-hook-install.mdmarkdown9.2 KB
catalog.mdmarkdown10.3 KB
ctx-fit/ARCHITECTURE_CANDIDATES.mdmarkdown12.9 KB
ctx-fit/ARCHITECTURE_GAP_ANALYSIS.mdmarkdown10.7 KB
ctx-fit/CTX_AUDIT.mdmarkdown11.6 KB
ctx-fit/CURRENT_STATE.mdmarkdown16.3 KB
ctx-fit/DECISIONS.mdmarkdown15.7 KB
ctx-fit/GOOD_FIRST_ISSUES.mdmarkdown20.6 KB
ctx-fit/IMPLEMENTATION_TRACKER.mdmarkdown697 B
ctx-fit/MAP.mdmarkdown7.8 KB
ctx-fit/PRD.mdmarkdown9.7 KB
ctx-fit/STATE.mdmarkdown60.9 KB
ctx-fit/milestones/M1_SPIKE_FRICTION_LOG.mdmarkdown5.5 KB
ctx-fit/milestones/MILESTONE_00_AUDIT.mdmarkdown5.1 KB
ctx-fit/planned_tasks.mdmarkdown17.7 KB
dashboard.mdmarkdown24.7 KB
enterprise-readiness-review.mdmarkdown8.7 KB
entity-onboarding.mdmarkdown18.0 KB
harness/attaching-to-hosts.mdmarkdown16.2 KB
harness/clean-host-contract.mdmarkdown4.5 KB
harness/loopflow-adapter-demo.mdmarkdown11.1 KB
huggingface-publish.mdmarkdown3.3 KB
index.mdmarkdown22.4 KB
knowledge-graph.mdmarkdown22.7 KB
marketplace-registry.mdmarkdown4.0 KB
memory-anchor.mdmarkdown2.7 KB
plans/ctx-agent-reviewer-workbench.mdmarkdown12.4 KB
plans/unified-capability-engine.mdmarkdown59.0 KB
qa/dashboard-user-story-status.csvother26.0 KB
qa/feature-user-story-status.csvother92.5 KB
research/context-engineering-guidance-audit.mdmarkdown10.4 KB
services/macos/com.claude.backup.watchdog.plistother1.7 KB
services/systemd/claude-backup-watchdog.serviceother1.3 KB
skill-lifecycle-and-dashboard.mdmarkdown6.2 KB
skill-quality-install.mdmarkdown6.2 KB
skill-router/index.mdmarkdown2.5 KB
skill-stack-matrix.mdmarkdown5.5 KB
skills-health.mdmarkdown3.5 KB
stack-signatures.mdmarkdown6.2 KB
telemetry.mdmarkdown24.7 KB
threat-model.mdmarkdown15.0 KB
toolbox/behavior-miner.mdmarkdown2.6 KB
toolbox/configuration.mdmarkdown4.0 KB
toolbox/council-runner.mdmarkdown2.4 KB
toolbox/hooks.mdmarkdown2.6 KB
toolbox/index.mdmarkdown3.0 KB
toolbox/intent-interview.mdmarkdown3.0 KB
toolbox/starters.mdmarkdown3.0 KB
toolbox/templates/docs-review.jsonconfig599 B
toolbox/templates/fresh-repo-init.jsonconfig533 B
toolbox/templates/refactor-safety.jsonconfig634 B
toolbox/templates/security-sweep.jsonconfig636 B
toolbox/templates/ship-it.jsonconfig676 B
toolbox/verdicts.mdmarkdown4.0 KB

toolbox/configuration.md

View on GitHub
# Configuration

Toolbox config lives in two files:

| Layer | Path | Format | Scope |
|---|---|---|---|
| **Global** | `~/.claude/toolboxes.json` | JSON | Every repo on this machine |
| **Per-repo** | `.toolbox.yaml` (project root) | YAML | This repo only, overrides global |

Per-repo entries shadow global entries with the same name. Fields absent
from the per-repo file fall back to the global value.

## Schema

```jsonc
{
  "version": 1,
  "toolboxes": {
    "<name>": {
      "description": "human-readable purpose",

      // Skills to load before the trigger fires
      "pre": ["python-patterns", "docs-lookup"],

      // Agents to run after
      "post": [
        "code-reviewer",
        "security-reviewer",
        "architect-review"
      ],

      "scope": {
        // "diff" | "dynamic" | "graph-blast" | "full"
        "analysis": "dynamic",
        // Optional: restrict to these glob projects
        "projects": ["*"],
        // Optional: match intent signals
        "signals": ["python"]
      },

      "budget": {
        "max_tokens": 60000,
        "max_seconds": 180
      },

      "dedup": {
        // "fresh" = always re-run, "cached" = reuse a matching
        // plan within window_seconds
        "policy": "cached",
        "window_seconds": 3600
      },

      "trigger": {
        "slash": true,
        // Optional file-save glob; null disables file-save
        "file_save": null,
        "pre_commit": true,
        "session_end": false
      },

      // If true, HIGH/CRITICAL verdicts block pre-commit
      "guardrail": true
    }
  }
}
```

## Field reference

### `pre` and `post`

- `pre` — skills to load before work starts. Loaded into the session's
  skill manifest, unloaded when the session ends.
- `post` — agents to invoke after the trigger. Each runs in its own
  sub-agent context window.

Either list can be empty. A toolbox with only `pre` is a skill preloader;
one with only `post` is a review council.

### `scope.analysis`

Controls what files the council sees:

| Value | Behavior |
|---|---|
| `diff` | Only files with uncommitted changes. Cheapest, fastest. |
| `dynamic` | Diff by default; expands one graph hop for tiny diffs when graph edges are available; falls back to full when no diff exists. |
| `graph-blast` | Current diff plus one-hop graph expansion when a graph edge map is supplied; otherwise the changed set. |
| `full` | Every tracked file. Most thorough; expensive — reserve for security sweeps. |

### `budget`

Copied into the `RunPlan` as `budget_tokens` and `budget_seconds`.
Downstream council execution enforces those caps.

### `dedup`

`fresh` always builds a new plan. `cached` reuses a matching plan when its
deterministic `plan_hash` is still within `window_seconds`. Dedup state lives
at `~/.claude/toolbox-runs/<plan_hash>.json`.

### `trigger`

Multiple triggers are allowed — a `ship-it` toolbox typically enables
`slash`, `pre_commit`, and `session_end`. `file_save` is a glob string
such as `"**/*.md"`; use `null` to disable file-save matching.
`session-start` is not configured in the trigger map: any active toolbox
with a non-empty `pre` list can preload those skills at session start.

### `guardrail`

When `true` and the trigger is `pre_commit`, the hook reads
`<plan_hash>.verdict.json` after the council runs and exits `2` (blocks
the commit) if level is `HIGH` or `CRITICAL`. See
[Verdicts & guardrails](verdicts.md).

## Editing tools

```bash
# List all toolboxes, both layers merged
python -m toolbox list

# Show resolved config for one toolbox
python -m toolbox show ship-it

# Activate a starter preset
python -m toolbox activate ship-it

# Export merged config
python -m toolbox export > my-toolboxes.yaml

# Import from file
python -m toolbox import my-toolboxes.yaml
```

## Validation

`toolbox_config` validates on read:

- `version` must equal `1`.
- `scope.analysis` must be one of `diff`, `dynamic`, `graph-blast`, `full`.
- `dedup.policy` must be one of `fresh`, `cached`.
- `budget.max_tokens` and `budget.max_seconds` must be positive ints.

Invalid entries raise `ValueError` with the offending key.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall