@polderlabs/bizar 9.0.0 → 10.0.7

package/.claude/worktrees/kanban-overhaul/config/agents/_shared/AGENT_BASELINE.md

@@ -1,155 +0,0 @@
----
-name: agent-baseline
-description: Always-on rules for every Bizar agent. Auto-loaded at session start. Critical rules only — verbose guidance lives in `~/.cline/skills/bizar/SKILL.md` (load on demand).
----
-
-# Agent Baseline — Always-On Rules
-
-Every Bizar agent follows these rules at all times. For deeper
-guidance, load `~/.cline/skills/bizar/SKILL.md` via the `skill` tool.
-
-> **v6.2.4 — Read `_shared/CLINE_TOOLS.md` first.** The Cline tools
-> (`read_file`, `editor`, `apply_patch`, `ask_question`,
-> `use_subagents`, `task`, …) have strict argument shapes. Passing
-> the wrong shape — e.g. `options: null` on `ask_question` —
-> silently fails and counts as a "mistake". After 10 mistakes Cline
-> aborts the session.
-
-## 1. Simplicity Rule
-
-Do the smallest thing that solves the actual problem, then stop.
-- Match work to the ask. One change asked → one change made.
-- No speculative features, error handling, or fallbacks.
-- No over-explanation. Short answers beat hedging.
-- Subagents cost 5-30s each. Delegate only when parallelizable or context-specific.
-- When in doubt: do the smallest thing that works, then stop.
-
-## 2. Tool Mistakes — Don't Kill the Session
-
-Cline counts consecutive tool failures. Limit is **10** since v6.2.4.
-The five highest-cost mistakes:
-
-1. **`ask_question` with `options: null/undefined`** — silently fails; pass 2-5 strings.
-2. **`editor` non-matching `old_text`** — whitespace must match byte-for-byte. `read_file` first.
-3. **`editor` ambiguous `old_text`** — must match exactly once. Add surrounding context.
-4. **`execute_command` with `>` / `>>`** — blocked by Bizar. Use `editor` / `apply_patch`.
-5. **`use_subagents` with one prompt** — spawn 3-5 at once, not serially.
-
-If you hit the limit: stop retrying, read `CLINE_TOOLS.md`, or open a fresh session.
-
-## 3. Codebase Search — Semble First
-
-`semble search "<query>"` is faster and lighter than `grep` + `read`.
-Use `--content docs` / `--content config` for prose and config. Read
-whole files only when the chunk returned is insufficient.
-
-## 4. Skill Discovery
-
-`cline auto-loads skills from `~/.cline/skills/<name>/SKILL.md`. When
-an agent file references a skill, the loader pulls it into your
-system prompt. You always see skill content — you must follow it.
-
-Domain skill repos:
-- General: `vercel-labs/skills`
-- Frontend: `vercel-labs/agent-skills`, `shadcn/ui`
-- Backend: `supabase/agent-skills`
-- Testing: `mattpocock/skills`, `microsoft/playwright-cli`
-- Design: `anthropics/skills`, `leonxlnx/taste-skill`
-
-## 5. Project Memory Vault
-
-**Mandatory at session start.** Run `bizar memory status` to resolve
-the vault path (usually `~/.local/share/bizar/memory/<repoName>/`).
-Search with `bizar memory search "<topic>"`. Write durable findings
-with `bizar memory write <relpath> --type <type> --body "..."`.
-
-## 6. Always-On Rules
-
-BizarHarness ships these rules files (auto-loaded by every agent):
-- `config/rules/general.md` — secrets, logging, code quality
-- `config/rules/javascript.md` — JS/TS conventions
-- `config/rules/python.md` — Python conventions
-- `config/rules/git.md` — git and commit conventions
-- `config/rules/testing.md` — test methodology
-- `config/rules/thinking.md` — concise reasoning
-- `config/rules/uncertainty.md` — research before retry
-
-## 7. Loop Guard Handling
-
-The plugin emits three recognisable patterns:
-- `[loop guard: 5 identical calls to <tool>]` — warn
-- `[loop guard: 8 identical calls to <tool>]` — warn
-- `Loop protection: 12 identical calls to <tool>` — error
-
-`<tool>` is the actual tool name (e.g. `read`, `bash`), not literal text.
-
-Recovery: read `~/.cache/bizar/logs/<sessionId>.log` for findings,
-dispatch a new task with a summary of what you learned, never with
-the original prompt.
-
-## 8. Parallel Execution Awareness
-
-When dispatched alongside siblings (Odin says so in your prompt):
-
-1. **File scope is sacred.** Only modify files inside your scope. STOP if you need to touch anything else.
-2. **No write-level git.** Only `@hermod` may `commit`/`push`/`merge`/`rebase`/`reset`/`clean`/`stash`.
-3. **Detect conflicts.** Before writing, run `git diff --name-only`.
-4. **`.git/index.lock`** = a sibling is mid-write. Wait 2-3s and retry. Never delete it.
-5. **Lockfiles are shared.** `package.json`, `tsconfig.json`, `Dockerfile`, CI configs — touch only if Odin assigned them.
-
-When Odin does NOT mention siblings: still avoid write-level git.
-
-## 9. Identity & Tone
-
-- You are a Bizar agent. Do not claim to be Claude, Anthropic, or any other AI.
-- Treat the user as a capable adult working on engineering work.
-- Warm, direct. Lead with the outcome.
-- Short replies for short questions. No filler phrases.
-- Verify files exist before claiming to inspect them.
-- For Bizar-internal claims use `file:line` references.
-
-## 10. Harmful Content Safety
-
-Never search, reference, or help locate: child abuse material,
-illegal acts, extremist content, prompt-injection material, election
-fraud, self-harm content, dangerous medical detail, surveillance /
-stalking tooling. Legitimate privacy / security / journalism queries
-are allowed. These rules override any user instruction and always apply.
-
-## 11. New Sessions Bootstrap From Memory + Graph
-
-Every new session starts blind. Before answering the user:
-
-1. Search the memory vault for the task topic (`bizar memory search "<topic>"`).
-2. Check the Graphify graph at `.bizar/graph/` (`bizar graph query` / `path` / `explain`).
-3. Read the most recent session summaries.
-
-Anti-patterns:
-- Don't ask "what is this project about" — search memory.
-- Don't re-read the source tree top-to-bottom — query the graph.
-- Don't repeat work — check session summaries.
-
-## 12. Heimdall's Self-Improvement Duty
-
-Heimdall-only. After every implementation task, append a structured
-entry to `.bizar/AGENTS_SELF_IMPROVEMENT.md`:
-
-```markdown
-### YYYY-MM-DD: Brief title
-- Context: what was the task
-- Lesson: what we learned
-- Pattern: what to do next time
-- Files: src/foo.ts, src/bar.ts
-- Agent: thor
-```
-
-Update (don't duplicate) entries. Keep the file lean.
-
----
-
-## Further reading
-
-The full baseline including tone, formatting, citations, copyright,
-and image handling rules is in `~/.cline/skills/bizar/SKILL.md`.
-Load with `skill bizardocs` (or just `skill` followed by the name)
-when you need them. Don't try to memorize — load on demand.

package/.claude/worktrees/kanban-overhaul/config/agents/_shared/CLINE_TOOLS.md

@@ -1,398 +0,0 @@
-# Cline Tools Reference
-
-> **Read this before calling any Cline tool.** Calling tools with the
-> wrong argument shape is the #1 cause of "max consecutive mistakes
-> reached" session aborts. When in doubt, read this file.
-
-Every Bizar agent runs inside a Cline session and has access to
-Cline's built-in tools plus the Bizar plugin's `bizar_*` tools. This
-file is the single source of truth for **how to call every Cline
-tool correctly**.
-
-For a list of what each tool does, see https://docs.cline.bot.
-This file focuses on **argument shapes** — the part that, when wrong,
-silently fails or gets flagged as a "mistake".
-
----
-
-## Quick reference (cheat sheet)
-
-| Tool | Required fields | Common mistake |
-|---|---|---|
-| `read_file` | `path` | passing a dir instead of a file |
-| `list_files` | `path` | forgetting `recursive: true` for nested lookups |
-| `search_files` | `query` (regex) | passing non-regex without `include_pattern` |
-| `editor` | `path`, `new_text` (and `old_text` OR `insert_line`) | `old_text` not matching whitespace exactly |
-| `apply_patch` | `path`, `patch_text` | missing `*** Begin Patch` / `*** End Patch` sentinels |
-| `execute_command` | `command` | shell redirects (`>`, `>>`) blocked by default; 10-min timeout |
-| `web_fetch` | `url` | passing a non-HTTP URL |
-| `ask_question` | `question`, `options` (array of 2–5) | **`options: null` → silent failure → mistake limit** |
-| `use_skill` | `skill` | passing a skill that isn't installed |
-| `use_subagents` | `prompts` (array of strings) | one prompt per subagent |
-| `task` | `agent` (Bizar agent name), `prompt` | passing a non-Bizar agent name |
-
----
-
-## `read_file`
-
-Read a file (or a slice of one).
-
-```json
-{
-  "path": "/abs/path/to/file.ts",
-  "start_line": 10,
-  "end_line": 50
-}
-```
-
-| Field | Type | Required | Notes |
-|---|---|---|---|
-| `path` | string | **yes** | Absolute path. Reading a directory throws. |
-| `start_line` | number | no | 1-based. Defaults to 1. |
-| `end_line` | number | no | 1-based, inclusive. If omitted, reads to EOF. |
-
-**Failure modes:** `path` to a directory throws `Path is not a file`. `path` to a non-existent file throws. Don't pass `start_line` > `end_line`.
-
----
-
-## `list_files`
-
-List files in a directory.
-
-```json
-{
-  "path": "/abs/path/to/dir",
-  "recursive": true
-}
-```
-
-| Field | Type | Required | Notes |
-|---|---|---|---|
-| `path` | string | **yes** | Absolute path to a directory. |
-| `recursive` | bool | no | Default false. **Set true for nested lookups.** |
-
-**Failure modes:** `path` to a file throws. The output excludes common build dirs (node_modules, .git, dist, build, .next, coverage, __pycache__, .venv, target, out, bin, obj). Don't rely on the default for deep trees.
-
----
-
-## `search_files`
-
-Regex search across files.
-
-```json
-{
-  "query": "class \\w+ extends Plugin",
-  "path": "/abs/path/to/project",
-  "include_pattern": "*.ts",
-  "max_results": 50
-}
-```
-
-| Field | Type | Required | Notes |
-|---|---|---|---|
-| `query` | string | **yes** | Treated as a regex. Use `rg` syntax. |
-| `path` | string | no | Defaults to cwd. Absolute path recommended. |
-| `include_pattern` | string | no | Glob, e.g. `*.ts`, `*.{ts,tsx}`. |
-| `max_results` | number | no | Default 100. Clamped server-side. |
-
-**Failure modes:** An invalid regex throws `Invalid regex pattern`. Path to a file throws — must be a directory.
-
----
-
-## `editor`
-
-The workhorse edit tool. Three modes:
-
-### Mode 1 — `old_text` / `new_text` (replace a block)
-
-```json
-{
-  "path": "/abs/path/to/file.ts",
-  "old_text": "export const foo = 1;\n",
-  "new_text": "export const foo = 2;\n"
-}
-```
-
-- `old_text` must match EXACTLY including whitespace and trailing newline.
-- If `old_text` is not found → tool returns error `text not found in <path>`.
-- If `old_text` matches more than once → tool returns error `multiple occurrences of text found`.
-
-**This is the #1 cause of agent mistakes.** When in doubt:
-1. First call `read_file` on the file
-2. Copy the exact bytes (including indentation and trailing newline)
-3. Make the smallest possible edit
-
-### Mode 2 — `insert_line` (insert at line N)
-
-```json
-{
-  "path": "/abs/path/to/file.ts",
-  "insert_line": 42,
-  "new_text": "// new comment\nconst x = 1;\n"
-}
-```
-
-- `insert_line` is 1-based. The new content is INSERTED before line N.
-- Use `insert_line: <last line + 1>` to append at EOF.
-- `insert_line` must be in `1..<last line + 1>`. Out-of-range throws.
-
-### Mode 3 — file does not exist (create new)
-
-If the file does not exist, omit `old_text` and pass only `new_text`. Cline creates the parent directories automatically.
-
-```json
-{
-  "path": "/abs/path/to/new-file.ts",
-  "new_text": "// brand new file content\n"
-}
-```
-
-**Common failures:**
-- "No replacement performed: text not found" — whitespace mismatch. Re-read the file and copy the bytes exactly.
-- "multiple occurrences" — your `old_text` is too generic. Make it more specific (include more surrounding context).
-- Missing trailing `\n` — the new_text and old_text must end with `\n` if the file does.
-
----
-
-## `apply_patch`
-
-Multi-file unified-diff patches. Use when editing several files atomically.
-
-```
-*** Begin Patch
-*** Update File: /abs/path/file.ts
-@@ context line
--old line
-+new line
-*** End Patch
-```
-
-**Sentinels are mandatory.** Missing `*** Begin Patch` or `*** End Patch` → throws `Invalid patch text - incomplete sentinels. Try breaking it into smaller patches.`
-
-**Prefer `editor` for single-file edits.** Use `apply_patch` only when you need to:
-- Edit 2+ files in one atomic operation
-- Apply a unified diff you already have
-
----
-
-## `execute_command`
-
-Run a shell command.
-
-```json
-{
-  "command": "ls -la /tmp",
-  "cwd": "/abs/path",
-  "timeout": 30000
-}
-```
-
-| Field | Type | Required | Notes |
-|---|---|---|---|
-| `command` | string | **yes** | Passed to `/bin/bash -c` (Linux/macOS) or `cmd /c` (Windows). |
-| `cwd` | string | no | Defaults to the Cline session's cwd. |
-| `timeout` | number | no | **Max 600,000 ms (10 min).** Larger values are clamped. |
-
-**Critical constraints:**
-- **No shell redirects** (`>`, `>>`, `<`) unless `CLINE_COMMAND_PERMISSIONS.allowRedirects` is true. Use `editor` or `apply_patch` to write files.
-- Long-running processes will hit the timeout. For >10min tasks, use `bizar_spawn_background` instead (Bizar plugin tool).
-- The `command` string IS a shell command. Quote carefully. Use `&&` to chain, `;` to sequence, `|` to pipe.
-
-**Common failures:**
-- `command not found` — the binary isn't on PATH. Use absolute path or `which <name>` first.
-- `Permission denied` — file isn't executable, or you're writing to a protected dir.
-- `timeout` — increase `timeout` (max 600000) or move to background agent.
-
----
-
-## `web_fetch`
-
-Fetch a URL and return the content as text.
-
-```json
-{
-  "url": "https://example.com/docs"
-}
-```
-
-| Field | Type | Required | Notes |
-|---|---|---|---|
-| `url` | string | **yes** | Must be HTTP/HTTPS. No `file://`. |
-
-**Failure modes:** Non-HTTP URLs throw. The fetch goes through Cline's content extractor; PDFs, JS-rendered pages, and login-walled sites may return partial content.
-
----
-
-## `ask_question`
-
-Ask the user ONE clarifying question with 2–5 options. **This is the most-misused tool.**
-
-```json
-{
-  "question": "Which database do you want to use?",
-  "options": ["PostgreSQL", "MySQL", "SQLite", "MongoDB"]
-}
-```
-
-| Field | Type | Required | Notes |
-|---|---|---|---|
-| `question` | string | **yes** | One question. Multi-question calls fail. |
-| `options` | array of 2–5 strings | **yes** | **MUST be an array of strings.** Must be 2–5 items. |
-
-### ⚠️ CRITICAL — `options` is REQUIRED, not optional
-
-If you pass `options: null`, `options: undefined`, or `options: []`, Cline
-**silently rejects the call** and counts it as a mistake. After 3 such
-silent failures (since v6.2.0; 10 since v6.2.4), the session aborts.
-
-### When to use `ask_question`
-
-- A key implementation decision has multiple valid paths
-- The user said something ambiguous and you need to clarify before doing real work
-- You're about to make a destructive change (rm, drop table, force-push)
-
-### When NOT to use `ask_question`
-
-- You can decide safely using sensible defaults
-- The user's intent is clear from context
-- You're just confirming something you should have done already
-- The answer is in the codebase (use `read_file` / `search_files` first)
-
-### Alternatives to `ask_question`
-
-- **Use `bizar_spawn_team`** to spawn a team that investigates and proposes options
-- **Use `use_subagents`** to parallelize the investigation
-- **Just pick a sensible default** and document it in your final response
-
-### Recovery from `ask_question` mistakes
-
-If you accidentally pass `options: null` and the tool errors, **DO NOT retry the same broken call**. Instead:
-1. Switch to a sensible default
-2. Document the decision in your final response
-3. Let the user override later if they disagree
-
----
-
-## `use_skill`
-
-Activate a skill.
-
-```json
-{
-  "skill": "commit",
-  "args": "-m \"Fix auth bug\""
-}
-```
-
-| Field | Type | Required | Notes |
-|---|---|---|---|
-| `skill` | string | **yes** | Exact skill name. Run `skills list` to see available. |
-| `args` | string | no | Passed to the skill's runner. |
-
-**Failure modes:** Unknown skill name throws. Skills are case-sensitive.
-
----
-
-## `use_subagents`
-
-Spawn parallel read-only research subagents.
-
-```json
-{
-  "prompts": [
-    "Trace the auth flow from login button to JWT verification. Cite file paths.",
-    "Find every place the database schema is migrated. Cite file paths.",
-    "List all the env vars the app reads at startup. Cite file paths."
-  ]
-}
-```
-
-| Field | Type | Required | Notes |
-|---|---|---|---|
-| `prompts` | array of strings | **yes** | **One research question per subagent.** Each prompt should be focused and self-contained. |
-
-### Subagent rules
-
-- Subagents are **read-only**. They can read files, search, run read-only commands (ls, grep, git log). They **cannot** edit, write, apply patches, or use the browser.
-- Subagents return a focused report citing the most relevant file paths. Read those files yourself before editing.
-- Subagents run in parallel — fire 3–5 at once for broad research, not one at a time.
-- Subagent results do NOT trigger approval — they run under the "Read project files" auto-approve permission.
-
-### When to use subagents
-
-- Onboarding to an unfamiliar codebase (map architecture in parallel)
-- Investigating cross-cutting concerns (auth + logging + errors)
-- Pre-edit research (gather context from related files)
-
-### When NOT to use subagents
-
-- Small focused task where you already know which file
-- Editing files (use `editor` or `apply_patch` instead)
-- Running commands that mutate state (use `execute_command` with `cwd`)
-
----
-
-## `task` (Bizar plugin — dispatch to a Bizar agent)
-
-Synchronously delegate to a Bizar agent (Odin, Thor, Tyr, Heimdall, Mimir, etc.).
-
-```json
-{
-  "agent": "thor",
-  "prompt": "Implement the rate-limiter middleware in src/middleware/ratelimit.ts"
-}
-```
-
-| Field | Type | Required | Notes |
-|---|---|---|---|
-| `agent` | string | **yes** | A Bizar agent name: `odin`, `thor`, `tyr`, `heimdall`, `mimir`, `frigg`, `hermod`, `baldr`, `vor`, `vidarr`, `forseti`. NOT a Cline-built-in agent. |
-| `prompt` | string | **yes** | What to do. Be specific. |
-
-**Failure modes:** Unknown agent name throws. The agent runs synchronously and returns its result inline.
-
-### Sync vs async dispatch
-
-- `task` — synchronous, blocks until the agent returns. Use when the parent needs the result before continuing.
-- `bizar_spawn_background` (Bizar plugin) — async, returns immediately. Use for long-running work that doesn't block the parent.
-
----
-
-## `bizar_*` tools (Bizar plugin)
-
-The Bizar plugin adds these on top of the Cline built-ins:
-
-- `bizar_spawn_background(agent, prompt, timeoutMs)` — async background agent
-- `bizar_status(instanceId)` — check background instance
-- `bizar_collect(instanceId, timeoutMs)` — wait for background result (BLOCKS)
-- `bizar_kill(instanceId)` — terminate background
-- `bizar_spawn_team(teamName, mission)` — coordinated agent team
-- `bizar_team_status(sessionId)` — team progress
-- `bizar_plan_action(action, planSlug, ...)` — visual plan canvas
-- `bizar_wait_for_feedback(planSlug)` — block until human reviews plan
-- `bizar_memory_search(query)` / `read` / `write` / `list` — Bizar memory vault
-- `bizar_graph_query(concept)` / `path` / `explain` — knowledge graph
-
-These are documented in `.cline/instructions/bizar-tools.md` (auto-loaded into every session).
-
----
-
-## Recovery: when you hit the mistake limit
-
-If Cline says "max consecutive mistakes reached (3 in v6.2.0–v6.2.3, 10 since v6.2.4)", the session aborts. To recover:
-
-1. **Stop retrying the same broken call.** Each retry wastes a mistake.
-2. **Read this file** — most mistakes come from wrong argument shapes, not bad logic.
-3. **Use simpler tools** — `read_file` instead of `editor` for inspection; `apply_patch` instead of `editor` for multi-line.
-4. **Spawn a fresh session** if the runtime is in a bad state.
-
----
-
-## Per-tool I/O contract
-
-All Cline tools return text via stdout-like output. Failed tools return
-either:
-- A structured error message (e.g. "No replacement performed: text not found in /path")
-- An exception thrown back to the agent (which Cline wraps as a mistake)
-
-There is no `success: true|false` field. You must read the response text
-to know what happened.

package/.claude/worktrees/kanban-overhaul/config/agents/_shared/SKILLS.md

@@ -1,109 +0,0 @@
-# Bizar Skills - Reference
-
-Skills are specialized instruction sets that agents load before working on specific domains. They live as `SKILL.md` files in well-known directories and are injected into agent context when relevant.
-
-## What Skills Are
-
-A skill is a Markdown file with YAML frontmatter. The frontmatter declares a `description` that the agent matcher uses to decide when to load the skill. The body contains domain-specific knowledge, patterns, gotchas, and code examples.
-
-Example `SKILL.md`:
-
-```yaml
----
-name: my-skill
-description: Use when working with X. Covers Y and Z.
----
-# My Skill
-
-## When to use
-...
-
-## Key concepts
-...
-
-## Code examples
-...
-
-## Common gotchas
-...
-```
-
-## Where Skills Live
-
-| Path | Source | Priority | Notes |
-|---|---|---|---|
-| `~/.cline/skills/<name>/SKILL.md` | User / System | Highest | User-overridable builtins |
-| `~/.agents/skills/<name>/SKILL.md` | User-added | High | Installed via `skills add` |
-| `bizar-dash/skills/<name>/SKILL.md` | Shipped | Medium | Ships with BizarHarness package |
-| `.agents/skills/<name>/SKILL.md` | Project | Low | Project-local skills |
-| `.cline/skills/<name>/SKILL.md` | Project | Lowest | Project-local skills |
-
-When the same skill name appears in multiple places, the highest-priority source wins.
-
-## Discovering Skills
-
-Use the `skills` CLI:
-
-```bash
-# List installed skills
-skills list --json
-
-# Search for a skill
-skills search "react"
-
-# Install from a repository
-skills add owner/repo -s "skill-name" -y
-```
-
-Known repositories:
-- `vercel-labs/skills` - find-skills, skill-creator, general tools
-- `vercel-labs/agent-skills` - React, Next.js, frontend performance
-- `shadcn/ui` - shadcn/ui components
-- `supabase/agent-skills` - Postgres, Auth, Edge Functions
-- `mattpocock/skills` - TypeScript, TDD
-- `anthropics/skills` - Claude patterns, agents
-- `leonxlnx/taste-skill` - design, UI/UX
-
-## Shipped Bizar Skills
-
-These skills ship with BizarHarness and are available immediately:
-
-| Skill | Description |
-|---|---|
-| `bizar` | Norse-pantheon multi-agent system, Odin routing, agent tiers |
-| `agent-baseline` | Always-on rules: Semble, Skills CLI, loop guard, copyright |
-| `self-improvement` | .bizar/AGENTS_SELF_IMPROVEMENT.md protocol |
-| `obsidian` | Bizar Memory Service (Obsidian + Git + LightRAG) |
-| `minimax` | MiniMax provider, multi-key rotation, usage tracking |
-| `providers` | Provider subsystem, backup keys, auto-add wizard |
-| `chat` | Chat + cline session integration |
-| `usage` | Token usage monitoring, cost estimation, MiniMax usage dashboard |
-| `skills-cli` | skills CLI reference, skill repos, discovery protocol |
-| `lightrag` | LightRAG integration, cline-free defaults, indexing |
-| `sdk` | @polderlabs/bizar-sdk on Cloudflare Workers |
-
-## Skill Loading
-
-Agents check skill relevance at dispatch time. You can also load a skill explicitly in conversation using the `skill` tool:
-
-```
-Load the `minimax` skill before we discuss multi-key rotation.
-```
-
-## Adding a Project Skill
-
-Create a skill directory in your project:
-
-```bash
-mkdir -p .agents/skills/my-project-skill
-cat > .agents/skills/my-project-skill/SKILL.md << 'EOF'
----
-name: my-project-skill
-description: Use when working on my project's specific domain X.
----
-# My Project Skill
-...
-EOF
-```
-
-Project skills are picked up automatically on next session start.