xonecas.io
64875
Mc

mini-coder / commit 31f7968

chore: cleanup repo skills

xonecas2026-03-31 31f7968c3cfde14361ebc249f508edf33bda1651 parent ec88c17

4 files changed

.agents/skills/audit/SKILL.md+0−63
@@ -1,63 +0,0 @@
1----
2-name: audit
3-description: "Full codebase audit — code review, multi-provider UI testing, and combined report. Use when the user asks to audit, test across providers, or validate the full implementation. Triggers on: audit, full review, test providers, validate implementation."
4-compatibility: "Requires tmux, bun, git, and active provider credentials (OpenAI OAuth and/or Opencode zen)"
5----
6-
7-Run a complete audit of the mini-coder implementation.
8-
9-## Phase 1 — Code & architecture review
10-
11-Audit the codebase against the project's core idea (the source of truth for design):
12-
13-1. Read the core idea section in AGENTS.md (or the system prompt context). And the KNOWN_ISSUES.md files.
14-2. Walk through each module and question design choices — stay compliant with the original idea.
15-3. Employ KISS, DRY, and YAGNI — flag surprise additions or eager optimizations.
16-4. Check `TODO.md` for known issues already tracked — don't duplicate them.
17-5. Verify documentation matches the idea file.
18-
19-Use `bun run dev` and `tmux` via shell to run one-shot and interactive sessions manually. Analyse console output and UI from a user's perspective. Follow the **tmux send-keys rules** in Phase 2 whenever interacting with tmux.
20-
21-## Phase 2 — Multi-provider UI probe
22-
23-Run `mc` sessions across all SDK paths to capture what real users see.
24-Do NOT delegate analysis to the models under test.
25-Use `tmux` so you can test interactive sessions properly.
26-
27-**tmux rules** (always follow these):
28-
29-- **Session naming**: always prefix tmux session names with `mc-audit-` (e.g. `mc-audit-haiku`, `mc-audit-gemini`). Never use generic names like `audit`, `test`, or `s` — they may collide with the host tmux session and kill the running mc process.
30-- Always use `-l` when sending text: `tmux send-keys -t <session> -l 'your text here'`
31-- Send `Enter` as a **separate** call without `-l`: `tmux send-keys -t <session> Enter`
32-- Never combine text and `Enter` in a single `send-keys` call — without `-l`, words like `Enter`, `Escape`, `Tab`, `Space` get interpreted as key presses instead of literal text.
33-- Pattern: `tmux send-keys -t mc-audit-test -l 'text'` then `tmux send-keys -t mc-audit-test Enter`
34-- Before creating a session, never run `tmux kill-session` or `tmux kill-server` — just use unique names.
35-
36-### Models to use (one per SDK path)
37-
38-Opencode zen and Openai login are available in this environment.
39-Pick cheap options. Verify available models at https://opencode.ai/docs/zen/, then choose:
40-
41-- `@ai-sdk/anthropic` path → a `zen/claude-*` haiku-class model
42-- `@ai-sdk/google` path → a `zen/gemini-*` flash-class model
43-- `@ai-sdk/openai` path (responses endpoint) → a `zen/gpt-*` small low-cost model
44-- `@ai-sdk/openai-compatible` path → `zen/*` or similar small low-cost model
45-
46-Avoid free models due to rate limit issues.
47-Also use an Openai OAuth model for completeness.
48-
49-### Prompts
50-
51-Use prompts that exercise realistic multi-step, multi-turn patterns. Check the db for examples if needed.
52-
53-## Phase 3 — Report
54-
55-Write `AUDIT-REPORT.md` in the repo root with these sections:
56-
57-- **Summary** — how you tested and what you found.
58-- **Code & Architecture** — correctness issues, KISS/DRY/YAGNI violations, idea misalignment.
59-- **UI/UX Alignment** — comparison of actual output vs the core idea's expectations across providers.
60-- **Recommendations** — split into immediate bugs, code changes, and polish items.
61-
62-Don't just recite the todo or known issues items back to the user — report on the audit itself and new findings.
63-Be concise. The report is analysis only — review findings with the user before making any changes.
.agents/skills/mini-coder/SKILL.md+0−201
@@ -1,201 +0,0 @@
1----
2-name: mini-coder
3-description: "How to use the `mc` CLI coding agent — interactive sessions, one-shot prompts, slash commands, file editing with mc-edit, and orchestrating subagents for complex tasks. Use when helping users operate mc, or when acting as an mc agent yourself."
4----
5-
6-# mini-coder
7-
8-Use this skill when you need to help a user operate the `mc` command-line tool, or when you are acting as an `mc` agent yourself.
9-
10-`mc` is the executable for **mini-coder**, a small CLI coding agent. This skill covers **how to use the tool**, not how the mini-coder repo is implemented.
11-
12-## Command-line usage
13-
14-```sh
15-mc # interactive session
16-mc "explain this codebase" # one-shot prompt
17-mc -c # continue most recent session
18-mc -r <session-id> # resume a specific session
19-mc -l # list recent sessions
20-mc -m <model-id> # pick a model
21-mc --cwd <path> # run in another directory
22-mc -h # help
23-```
24-
25-## Interactive slash commands
26-
27-Inside `mc`:
28-
29-| Command | Purpose |
30-| ----------------------------------------------- | --------------------------------------------- |
31-| `/help` | Show help |
32-| `/model` | List models |
33-| `/model <id>` | Switch model |
34-| `/model effort <low\|medium\|high\|xhigh\|off>` | Set reasoning effort |
35-| `/reasoning [on\|off]` | Toggle reasoning display |
36-| `/verbose` | Toggle output truncation |
37-| `/undo` | Remove last turn (does not revert filesystem) |
38-| `/new` | Start a fresh session |
39-| `/mcp list` | List MCP servers |
40-| `/mcp add <name> http <url>` | Add HTTP MCP server |
41-| `/mcp add <name> stdio <cmd> [args...]` | Add stdio MCP server |
42-| `/mcp remove <name>` | Remove MCP server |
43-| `/login [provider]` | Show or start OAuth login |
44-| `/logout <provider>` | Clear OAuth tokens |
45-| `/exit`, `/quit`, `/q` | Leave the session |
46-
47-## Prompt input features
48-
49-- **Shell input** — prefix with `!` to run a shell command inline and feed output to the LLM.
50-- **File references** — `@path/to/file` embeds a file into the prompt (autocompletes paths).
51-- **Skill / agent refs** — `/` triggers autocomplete for skills and agents.
52-- **Images** — paste image data URLs or file paths directly.
53-- **ESC** interrupts an in-progress response (preserves partial output). `Ctrl+C` exits forcefully, `Ctrl+D` exits gracefully.
54-
55-## File editing with mc-edit
56-
57-`mc` is shell-first. The preferred workflow is: **inspect → edit → verify**.
58-
59-```sh
60-mc-edit <path> (--old <text> | --old-file <path>) [--new <text> | --new-file <path>] [--cwd <path>]
61-```
62-
63-- Applies one exact-text replacement per invocation.
64-- Old text must match exactly once (fails on zero or multiple matches).
65-- Omit `--new` / `--new-file` to delete matched text.
66-- Prints a unified diff + metadata on success; errors go to stderr.
67-
68-### Examples
69-
70-```sh
71-mc-edit src/auth.ts --old "function login(user)" --new "async function login(user)"
72-
73-mc-edit src/config.ts --old "// TODO: remove this
74-const DEBUG = true;"
75-
76-mc-edit src/big.ts --old-file /tmp/old.txt --new-file /tmp/new.txt
77-```
78-
79-## Config locations
80-
81-mini-coder discovers configs from both `.agents` and `.claude` layouts:
82-
83-| Scope | Paths |
84-| ------ | -------------------------------------------------------------------------------------- |
85-| Local | `.agents/commands/*.md`, `.agents/agents/*.md`, `.agents/skills/<name>/SKILL.md` |
86-| Global | `~/.agents/commands/*.md`, `~/.agents/agents/*.md`, `~/.agents/skills/<name>/SKILL.md` |
87-
88-Same structure applies under `.claude/` and `~/.claude/`.
89-
90-App data (sessions, MCP config, logs) lives in `~/.config/mini-coder/`.
91-
92-## Orchestration mode
93-
94-For complex, multi-part tasks you can use `mc` as a subagent coordinator.
95-
96-### Core principles
97-
98-- **Decompose first.** Break the task into 2–5 independent, self-contained subtasks.
99-- **Delegate aggressively.** Spawn focused subagents via `mc "prompt"` in the shell tool — each gets a single precise goal.
100-- **Parallelise when possible.** Dispatch independent subtasks concurrently.
101-- **Stay the coordinator.** You own the big picture; subagents handle the details. Synthesise their results, resolve conflicts, report back clearly.
102-- **Fail fast.** If a subagent fails, stop and reassess before continuing.
103-
104-### Subagent prompt hygiene
105-
106-- Be explicit about what the agent should produce (a file, a code change, a summary).
107-- Include all relevant paths, constraints, and acceptance criteria — subagents have no shared memory.
108-- Never give vague briefs; ambiguity leads to rework and wasted tokens.
109-
110-### When NOT to delegate
111-
112-- Trivial lookups or single-file reads — do them directly.
113-- Tasks requiring continuous back-and-forth with the user — handle interactively.
114-
115-## Using tmux with mc
116-
117-When testing `mc` interactively from an outer agent or script, use `tmux` to drive the session. This avoids the limitations of non-interactive shell invocations.
118-
119-### Setup
120-
121-```sh
122-tmux new-session -d -s test -x 200 -y 50
123-tmux send-keys -t test 'bun run dev' Enter # or 'mc'
124-sleep 3 # wait for banner
125-```
126-
127-### Sending input
128-
129-Use `-l` (literal) to avoid key-name interpretation, then send `Enter` separately:
130-
131-```sh
132-tmux send-keys -t test -l "What files are in src/"
133-tmux send-keys -t test Enter
134-sleep 10
135-tmux capture-pane -t test -p -S -30 # read last 30 lines
136-```
137-
138-### Paste detection
139-
140-`mc` uses bracketed paste mode. When `tmux send-keys` delivers text rapidly (even with `-l`), the terminal may wrap it in paste brackets. This causes `mc` to show a `[pasted: "…"]` label instead of echoing the typed text. **This is expected behavior** — real users typing character-by-character won't trigger it.
141-
142-To work around it in automation:
143-
144-- **Short prompts**: use `send-keys -l "text"` + `send-keys Enter` — paste label is cosmetic, submission still works.
145-- **Slash commands**: same pattern — `/model zen/claude-haiku-4-5` + `Enter`.
146-- **Multi-line input**: not supported in the prompt; send one logical prompt per submission.
147-
148-### Reading output
149-
150-```sh
151-tmux capture-pane -t test -p -S -50 # last 50 lines
152-tmux capture-pane -t test -p -S -50 -E -1 # specific range
153-```
154-
155-Capture after a generous `sleep` — tool calls and streaming take time.
156-
157-### Sending special keys
158-
159-```sh
160-tmux send-keys -t test Escape # interrupt response (ESC)
161-tmux send-keys -t test C-d # graceful exit (Ctrl+D)
162-tmux send-keys -t test C-c # force exit (Ctrl+C)
163-```
164-
165-### Cleanup
166-
167-```sh
168-tmux send-keys -t test C-d
169-sleep 1
170-tmux kill-session -t test
171-```
172-
173-### Full example: test a model switch + prompt
174-
175-```sh
176-tmux new-session -d -s test -x 200 -y 50
177-tmux send-keys -t test 'mc' Enter
178-sleep 3
179-
180-tmux send-keys -t test -l "/model zen/gemini-3-flash"
181-tmux send-keys -t test Enter
182-sleep 2
183-
184-tmux send-keys -t test -l "List the files in the current directory"
185-tmux send-keys -t test Enter
186-sleep 15
187-
188-tmux capture-pane -t test -p -S -30 # review output
189-
190-tmux send-keys -t test C-d
191-sleep 1
192-tmux kill-session -t test
193-```
194-
195-## Tips for helping users
196-
197-- Prefer exact commands the user can paste.
198-- Distinguish **shell commands** (outside `mc`) from **slash commands** (inside `mc`).
199-- Suggest `mc -h` or `/help` for discovery.
200-- Suggest `mc -l` / `mc -c` / `mc -r <id>` for session recovery.
201-- Mention `--cwd` for wrong-directory issues, `-m` for model selection.
.agents/skills/release/SKILL.md+3−3
@@ -4,15 +4,15 @@ description: "Create a new tagged npm release (patch/minor/major). Use when the
44 compatibility: "Requires git, bun, and npm with publish credentials"
55 ---
66
7-Create a new npm release for this repo. The bump type is: $1 (default to "patch" if not provided).
7+Create a new npm release for this repo. Complete all steps in sequence.
88
99 ## Steps
1010
11-1. **Normalize** the bump type to one of `patch | minor | major`. Default to `patch` if empty.
11+1. **Normalize** the bump type to one of `patch | minor | major`. Default to `patch` if the user didn't specify.
1212 2. **Verify git state** before mutating anything:
1313 - Current branch must be `main`.
1414 - Working tree must be clean (`git status --porcelain` is empty).
15- - Run `git fetch origin --tags` and ensure `HEAD` equals `origin/main`.
15+ - Run `git fetch origin --tags` and ensure `HEAD` equals `origin/main`. Update the origin as needed.
1616 - If any check fails, stop and report.
1717 3. **Run checks** — execute the full repo check suite. Stop and report on failures.
1818 4. **Read `package.json`** and capture `name` and current `version`.
AGENTS.md+1−2
@@ -17,8 +17,7 @@
1717 - If you make temp files, clean them up when you are done.
1818 - Before committing, run `git status` and ensure every modified and untracked file produced during the session is included. Do not leave files behind.
1919 - Use Conventional Commits formatting for commit messages.
20-- Before committing code changes, review the diff with the user and get explicit approval for the commit. Do not treat this as a requirement to pause before every git command in a requested workflow. When you are allowed to commit, include the whole diff unless told otherwise.
21-- Before starting work, check superpowers skills (`listSkills`) and load any that match the task.
20+- Before committing code changes, review the diff with the user and get approval for the commit. Treat direct user requests to commit or to do repository tasks as approval.
2221
2322 ## Core idea, treat this as the source of truth for the design/implementation.