skill-system-comms
Unified agent communication and session lifecycle skill for tmux-routed reviewer messaging, startup soul-profile delivery, guarded dispatch-down flows, and canonical coder/user/auto session management.
What this skill does
# Skill System Comms
This skill is the merged owner of reviewer communication and coder lifecycle management.
It replaces manual reviewer tmux messaging, ad-hoc inbox files, and split session-lifecycle ownership with one structured command surface.
## Commands
```bash
bash "Work/.tkt/scripts/tmux-send-atom.sh" --key C-m <tmux-pane> <message>
bash "Work/.tkt/scripts/tmux-send-atom.sh" <tmux-pane> "/compact"
bash "Work/.tkt/scripts/tmux-send-atom.sh" --key Tab <tmux-pane>
bash "Work/.tkt/scripts/callosum.sh" render --meeting <meeting.md>
bash "Work/.tkt/scripts/callosum.sh" say --meeting <meeting.md> --speaker <wren|finch|lark> --message <message>
python3 "{skill_dir}/scripts/reviewer_comms.py" send <target> <message>
python3 "{skill_dir}/scripts/reviewer_comms.py" broadcast <message>
python3 "{skill_dir}/scripts/reviewer_comms.py" ack <message-id>
python3 "{skill_dir}/scripts/reviewer_comms.py" poll
python3 "{skill_dir}/scripts/reviewer_comms.py" meeting start <topic>
python3 "{skill_dir}/scripts/reviewer_comms.py" meeting ack <meeting-id> <reviewer>
python3 "{skill_dir}/scripts/reviewer_comms.py" meeting status <meeting-id>
python3 "{skill_dir}/scripts/reviewer_comms.py" meeting close <meeting-id>
python3 "{skill_dir}/scripts/callosum.py" ensure-board [--board dev_user:callosum]
python3 "{skill_dir}/scripts/callosum.py" render --meeting <meeting.md> [--board dev_user:callosum]
python3 "{skill_dir}/scripts/callosum.py" say --meeting <meeting.md> --speaker <wren|finch|lark> --message <message>
python3 "{skill_dir}/scripts/reviewer_comms.py" dispatch-down <project> <session-type> <message> [--notify] [--escalation-level L0|L1|L2|L3] [--force --reason <text>]
python3 "{skill_dir}/scripts/reviewer_comms.py" startup <reviewer> [--message <context>]
python3 "{skill_dir}/scripts/reviewer_comms.py" session-report scan <project-dir>
python3 "{skill_dir}/scripts/coder_lifecycle.py" discover <project-dir>
python3 "{skill_dir}/scripts/coder_lifecycle.py" ensure <project-dir>
python3 "{skill_dir}/scripts/coder_lifecycle.py" cleanup <project-dir>
python3 "{skill_dir}/scripts/coder_lifecycle.py" restart <project-dir> <session-name>
python3 "{skill_dir}/scripts/coder_lifecycle.py" audit
bash "{skill_dir}/scripts/coder/dispatch-coder.sh" <tmux-pane> <message>
bash "{skill_dir}/scripts/coder/restart-coder.sh" <tmux-pane> [--session <id>] [--workdir <path>]
bash "{skill_dir}/scripts/coder/capture-coder.sh" <tmux-pane> [lines] [--todo]
bash "{skill_dir}/scripts/coder/health-check-coder.sh" [tmux-pane]
bash "{skill_dir}/scripts/coder/bootstrap-coders.sh" [SK|FD|EP ...]
bash "{skill_dir}/scripts/coder/coder-sessions.sh" [--get <keyword>]
```
## Public CLI Wrapper
The thin public wrapper for coder harness commands is `sk coder`.
The original shell scripts remain valid for backward compatibility.
```bash
python3 "{skill_dir}/../skill-system-cli/scripts/sk.py" coder dispatch dev_coder:SK-coder "B-069 TKT-001: implement wrapper"
python3 "{skill_dir}/../skill-system-cli/scripts/sk.py" coder capture dev_coder:SK-coder --lines 80 --todo
python3 "{skill_dir}/../skill-system-cli/scripts/sk.py" coder restart dev_coder:SK-coder --session ses_demo --workdir /tmp/project
python3 "{skill_dir}/../skill-system-cli/scripts/sk.py" coder sessions --get skills
python3 "{skill_dir}/../skill-system-cli/scripts/sk.py" coder health dev_coder:SK-coder
python3 "{skill_dir}/../skill-system-cli/scripts/sk.py" coder bootstrap SK FD
```
## Rules
- tmux routing is by window name under `dev-user:<target>`.
- Message and ack artifacts are file-backed under `.tkt/`.
- Meeting close emits per-owner inbox YAML under `.tkt/inbox/<reviewer>/`.
- Same-level meetings use the corpus callosum path: `callosum.py say` appends the utterance to the meeting artifact, then renders the recent live log to `dev_user:callosum`.
- `callosum.py render --meeting <meeting.md>` refreshes the board without adding a new utterance.
- Editing a meeting artifact without a matching callosum live utterance is `callosum bypass`; acceptable for archival cleanup, not for active sister meeting turns.
- `callosum.py ensure-board` also attaches a pipe-pane audit log at `.tkt/logs/callosum-pane.log`; tmux scrollback is not durable evidence.
- Session-report reuses B-054 coder lifecycle discovery logic.
- `tmux-send-atom.sh` is the lowest-level tmux motor nerve: paste text and/or caller-specified key sequence only. Higher-level scripts compose it instead of open-coding `load-buffer` / `paste-buffer` / `send-keys`.
- `compact-session.sh` owns Codex/OpenCode self-compact gating: empty input + non-working pane + stable idle window, then bare `/compact` followed by independent `C-m`.
- dispatch-down rejects placeholder payloads and reports TASK/EXPECTED OUTCOME/CONTEXT quality.
- dispatch-down supports escalation levels L0-L3; L4 is intentionally unsupported.
- L1 prepends a status-check prefix without interrupting the coder pane.
- L2 invokes `Work/.tkt/scripts/force-report-coder.sh`, waits, then dispatches with an interruption prefix.
- L3 writes Hermes incident yaml + inbox artifacts before dispatching an emergency-prefixed message.
- Every L>=1 dispatch appends incident metadata under `Work/.tkt/incidents/hermes/` and enforces level-specific cooldowns unless `--force --reason` is supplied.
- dispatch-down triggers architecture escalation metadata after three retryable failures.
- startup loads reviewer soul profile mappings from `config/reviewer-comms.yaml`.
- canonical coder/user/auto session discovery and mutation live in `coder_lifecycle.py` under this same skill.
- coder harness assets live under `scripts/coder/` and remain reachable from `Work/.tkt/scripts/` via symlink compatibility.
Example escalation calls:
```bash
python3 "{skill_dir}/scripts/reviewer_comms.py" dispatch-down ep coder "B-039 status?" --escalation-level L1
python3 "{skill_dir}/scripts/reviewer_comms.py" dispatch-down sk coder "force surface SK-C blocker" --escalation-level L2
python3 "{skill_dir}/scripts/reviewer_comms.py" dispatch-down fd coder "EMERGENCY: stop B-076 deploy" --escalation-level L3 --force --reason "production cluster freeze in flight"
python3 "{skill_dir}/scripts/reviewer_comms.py" dispatch-down sk coder "self test" --self-test
```
Cooldown windows:
- `L1`: max 1 invocation per target per 30 minutes
- `L2`: max 1 invocation per target per 60 minutes
- `L3`: max 1 invocation per target per 24 hours and requires prior `L1`/`L2` attempts unless bypassed
- `--force` bypasses cooldown only when paired with `--reason "..."`, and the bypass is logged in the incident yaml
Intentional limit:
- `L4` is not implemented. The escalation ladder stops at `L3` by design.
```skill-manifest
{
"schema_version": "2.0",
"id": "skill-system-comms",
"version": "1.1.0",
"capabilities": ["reviewer-send", "reviewer-broadcast", "reviewer-ack", "reviewer-poll", "reviewer-meeting", "reviewer-callosum", "reviewer-dispatch-down", "reviewer-startup", "reviewer-session-report", "session-discover", "session-ensure", "session-cleanup", "session-restart", "session-audit", "coder-harness"],
"effects": ["fs.read", "fs.write", "proc.exec", "db.read"],
"operations": {
"send": {
"description": "Send a reviewer message to one named reviewer window and log it.",
"input": {
"target": {"type": "string", "required": true, "description": "Reviewer window target"},
"message": {"type": "string", "required": true, "description": "Message body"}
},
"output": {
"description": "Delivery result and message id",
"fields": {"message_id": "string", "delivery": "object"}
},
"entrypoints": {
"unix": ["python3", "{skill_dir}/scripts/reviewer_comms.py", "send", "{target}", "{message}"]
}
},
"broadcast": {
"description": "Send a reviewer message to all known reviewer windows.",
"input": {
"message": {"type": "string", "required": true, "description": "Message body"}
Related in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.