Dev Tier Tools (6)
These 6 tools require a midos_sk_dev_* API key or higher. They unlock full, untruncated knowledge access and session management.
get_skill
Section titled “get_skill”Get a specific skill document by name.
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
name | string | Yes | — | Skill name (case-insensitive, fuzzy matched) |
section | string | No | "" | Extract specific section by header |
page | int | No | 0 | Page number for pagination |
mode | string | No | "" | Set to "toc" for table of contents only |
Returns: Full skill document in markdown (up to 10,000 chars) or “not found.”
{ "name": "get_skill", "arguments": { "name": "fastapi-patterns", "section": "Dependency Injection" }}Notes: Fuzzy matching means "react" will find "react-comprehensive" or "react-19-migration". Use mode: "toc" to browse sections before requesting a specific one.
get_protocol
Section titled “get_protocol”Get an operational protocol or procedure document.
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
name | string | Yes | — | Protocol name (case-insensitive) |
Returns: Full protocol document in markdown (up to 10,000 chars) or “not found.”
chunk_code
Section titled “chunk_code”Parse a code file into semantic chunks for RAG ingestion.
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
file_path | string | Yes | — | Path to code file (within knowledge/) |
Returns: AST-based chunks with metadata including function signatures, class names, line numbers, and docstrings. Language auto-detected from file extension.
{ "name": "chunk_code", "arguments": { "file_path": "knowledge/chunks/example_patterns.py" }}memory_stats
Section titled “memory_stats”Get vector store diagnostics and statistics.
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
| (none) | — | — | — | — |
Returns: { total_chunks, cache_size, lance_db_status, embedding_model, cache_hit_rate }
Useful for monitoring vector store health and cache performance.
pool_status
Section titled “pool_status”Check multi-instance coordination pool status.
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
| (none) | — | — | — | — |
Returns: Current pool state showing active instances, claimed resources, and signals. Read-only — no writes allowed through this tool.
save_progress
Section titled “save_progress”Save a session checkpoint for later resumption via where_was_i or get_progress.
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
session_id | string | Yes | — | Unique session identifier |
summary | string | Yes | — | What was accomplished |
files_touched | string | No | "" | Comma-separated list of modified files |
decisions_made | string | No | "" | Key decisions during session |
next_steps | string | No | "" | What to do next time |
round_id | string | No | "" | Operation round identifier |
Returns: { status: "saved", session_id, round_id }
{ "name": "save_progress", "arguments": { "session_id": "api-refactor-2026-02", "summary": "Refactored auth middleware to use JWT. Tests passing.", "files_touched": "auth.py, middleware.py, tests/test_auth.py", "decisions_made": "Chose JWT over session cookies for stateless scaling", "next_steps": "Add refresh token rotation, update API docs" }}