Skip to main content
Glama
JeanFrancoisGagne

crapkit

by JeanFrancoisGagne
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Config and repo health check

check_config
Read-onlyIdempotent

Validate crapkit.toml against the repo to catch typo keys, empty scopes, missing lane working dirs, and runners that fail to start. Run it before trusting strange tool results or missing rankings.

Instructions

Checks that crapkit.toml agrees with the repo: typo keys, empty scopes, missing lane cwds, runners that fail to start. Run it first when any tool answers strangely or the ranking misses a file, and list_runs when only the history is in question. It needs no run, probes each runner once per lane directory and environment, runs no lane, and any problem arrives with isError true. repo can be any directory under the checkout, and one with no crapkit.toml above it answers a pointer, never a parent's config.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
repoNopath to the scored repo's root (default: the repo the server was started in); a leading ~ is the home directory

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault
lanesNoper declared lane
storeNothe run store
schemaNopayload schema version, 1
problemsNoFAIL findings as sentences naming the fix; non-empty is exit 1 and the result carries isError true
versionsNothe tools behind every number
warningsNoWARN findings: unmeasured directories, scopes with a lane but no scoped_tests template, artifacts written outside .crapkit, lanes without results_artifact; exit stays 0
resourcesNoeffective resource policy; limits and estimates, not sampled utilization
truncatedNopresent only when the whole answer was longer than one tool result the server sends (7,500 characters): each list, string or object field it cut, at any depth, keeps its start, and full is the CLI command that prints the whole answer
newest_runNothe newest run, or null when nothing has run
analysis_versionNothe analysis semantics version; with the lizard version it stamps every ratchet mark, so a bump refuses old marks until the repo re-seeds

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed3 schema fields changedv0.8.1
    • changedInput schema / properties / repo / description
      Previous value: -"path to the scored repo's root (default: the repo the server was started in)"New value: +"path to the scored repo's root (default: the repo the server was started in); a leading ~ is the home directory"
    • addedOutput schema / properties / lanes / items / properties / refusal
      Added value: +{
+  "description": "why --reuse-artifacts will not score the lane's artifact on disk: the lane's last attempt wrote no artifact and the file predates it, or .crapkit/artifacts.json cannot be read; null when reuse would score it",
+  "type": [
+    "string",
+    "null"
+  ]
+}
    • addedOutput schema / properties / truncated
      Added value: +{
+  "description": "present only when the whole answer was longer than one tool result the server sends (7,500 characters): each list, string or object field it cut, at any depth, keeps its start, and full is the CLI command that prints the whole answer",
+  "properties": {
+    "fields": {
+      "additionalProperties": {
+        "properties": {
+          "kept": {
+            "description": "how many the answer keeps",
+            "type": "integer"
+          },
+          "of": {
+            "description": "how many the whole answer has",
+            "type": "integer"
+          }
+        },
+        "type": "object"
+      },
+      "description": "each cut field, named by its keys joined with dots (gate.breaches) -> kept and of: elements for a list, entries for an object, characters for a string",
+      "type": "object"
+    },
+    "full": {
+      "description": "the crapkit command that prints the whole answer, spelled for the server's shell",
+      "type": "string"
+    }
+  },
+  "type": "object"
+}
  2. Changed2 schema fields changedv0.8.0
    • changedOutput schema / properties / resources / properties / test_retention_count / description
      Previous value: -"default test evidence count limit; zero disables it"New value: +"deprecated, always 0: crapkit applies no test evidence retention"
    • changedOutput schema / properties / resources / properties / test_retention_days / description
      Previous value: -"default test evidence age limit; zero disables it"New value: +"deprecated, always 0: crapkit applies no test evidence retention"
  3. Changed1 schema field changedv0.7.2
    • addedOutput schema / properties / resources
      Added value: +{
+  "description": "effective resource policy; limits and estimates, not sampled utilization",
+  "properties": {
+    "available_cpus": {
+      "description": "CPUs visible to this process",
+      "type": "integer"
+    },
+    "budget_directory": {
+      "description": "coordination directory; status does not create it",
+      "type": "string"
+    },
+    "coordination": {
+      "description": "worker slot ownership scope",
+      "type": "string"
+    },
+    "cpu_probe": {
+      "description": "CPU affinity or fallback probe used",
+      "type": "string"
+    },
+    "default_chunks_per_worker": {
+      "description": "chunk target used by automatic pool sizing",
+      "type": "integer"
+    },
+    "default_source_bytes_per_worker": {
+      "description": "source-byte target for automatic spawn sizing; null for other start methods",
+      "type": [
+        "integer",
+        "null"
+      ]
+    },
+    "estimated_pool_memory_mb": {
+      "description": "estimated memory for the allowed worker count",
+      "type": "integer"
+    },
+    "inherited_analysis_workers": {
+      "description": "valid inherited worker ceiling or null",
+      "type": [
+        "integer",
+        "null"
+      ]
+    },
+    "log_max_bytes": {
+      "description": "byte limit per current and backup lane log; zero is unlimited",
+      "type": "integer"
+    },
+    "memory_budget_mb": {
+      "description": "inherited memory sizing hint or null",
+      "type": [
+        "integer",
+        "null"
+      ]
+    },
+    "memory_is_hard_limit": {
+      "description": "false: memory policy sizes workers without an OS allocation limit",
+      "type": "boolean"
+    },
+    "pool_worker_limit": {
+      "description": "per-pool ceiling after CPU, memory and inherited limits",
+      "type": "integer"
+    },
+    "requested_analysis_workers": {
+      "description": "configured per-pool ceiling; zero selects the default",
+      "type": "integer"
+    },
+    "serial_fallback": {
+      "description": "busy slots allow serial work without waiting",
+      "type": "boolean"
+    },
+    "shared_pool_limit": {
+      "description": "shared numbered pool slot ceiling",
+      "type": "integer"
+    },
+    "test_retention_count": {
+      "description": "default test evidence count limit; zero disables it",
+      "type": "integer"
+    },
+    "test_retention_days": {
+      "description": "default test evidence age limit; zero disables it",
+      "type": "integer"
+    },
+    "worker_memory_estimate_mb": {
+      "description": "memory estimate per analysis worker",
+      "type": "integer"
+    }
+  },
+  "type": "object"
+}
  4. Addedv0.6.0

TDQS

A4.8/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations cover the safety profile (readOnly, idempotent, non-destructive), but the description adds behavior they cannot: no run is required, each runner is probed once per lane directory and environment, no lane is executed, and failures surface as isError true. That is real operational context an agent needs to interpret output.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with the purpose, then usage, then behavioral specifics; every sentence carries information. The third clause chains several facts together ('probes each runner once per lane directory and environment, runs no lane') which slightly hurts scannability but wastes no words.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

An output schema exists, so return-value detail is not required, yet the description still flags the isError signal and the no-config pointer case. For a single-optional-param diagnostic tool, nothing an agent needs to invoke or interpret it is missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100% and the single param is documented there, so the baseline is 3. The description goes beyond it by explaining resolution semantics: repo may be any directory under the checkout, and a path with no crapkit.toml above it returns a pointer rather than inheriting a parent's config.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Names a specific verb+resource ('Checks that crapkit.toml agrees with the repo') and enumerates the concrete failure classes it detects: typo keys, empty scopes, missing lane cwds, runners that fail to start. An agent can tell this apart from list_runs, get_trend, check_gate, etc. without opening any schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

'Run it first when any tool answers strangely or the ranking misses a file' gives an explicit trigger condition, and 'list_runs when only the history is in question' names the sibling and the condition that selects it. When-to-use and the alternative are both stated.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools