CodeEraser
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| CE_LANG | No | Output language for the server. English by default and Chinese under `--lang zh`, `CE_LANG=zh` or `[ui] lang = "zh"`. | en |
| CE_CORE_BIN | No | Path to the `ce-core` judgment core binary. `ce` finds the core as `CE_CORE_BIN` → a `ce-core` sibling on PATH → PATH; the `--core <path>` option wins over this. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| scanC | Size/complexity/readability metrics report (ce.scan-report schema). |
| check_duplicationC | Verified T1/T2 clone blocks (ce.dedup-report schema). |
| churnC | Append-vs-rewrite and co-change over a git window (ce.churn-report schema). |
| graph_sitesC | Reference sites, resolution-free (the ce graph --sites rows). |
| deadcodeC | Liveness verdicts over the reference graph, plus the symbol-level advisory of declarations no other file spells — never a verdict (ce.deadcode-report schema). |
| cloneC | T3 near-miss clone judgment (ce.clone-report schema); |
| docdupC | Documentation-duplication judgment (ce.docdup-report schema). |
| joinC | Three-signal join: similarity, graph position, churn (ce.join-report schema). |
| structureC | Tree-scale structure judgment, eight axes (ce.structure-report schema). |
| checkC | ADR-006 baseline judgment: score, ratchet and axes (report only — this surface never writes a baseline). |
| eraseA | The deterministic erase PLAN (ce.erase-plan schema) — dry-run by construction: this surface reaches the plan and nothing else, and applying it is a human act at the CLI or the GUI. |
| erase_logA | The applied-erase audit trail (ce.erase-trail-report schema): every record |
| doctorB | This machine's state (ce.doctor-report schema): ce-core handshake, the tier of every guard class the hooks read (the write guard, flow, tombstone), index freshness, daemon, degraded-run counter. The daemon is asked without being started. |
| trendB | Score trajectory over mainline git history (ce.trend-report schema; points cache in the index, rebuildable from history). |
| similar_unitsC | The same-role advisor (ce.similar-report schema): the units most like one unit or a text, in the core's order with its role bit and a six-channel evidence row; |
| queryB | One question in CE Datalog over the index's facts (ce.query-report schema): files, references, units, clones, mentions, path classes, with the built-in prelude ( |
| rulesB | The project's architecture rules judged (ce.rules-report schema): every |
| flowB | Dead code inside functions (ce.flow-report schema): unreachable statements, dead stores, unused locals and parameters, judged by the core over flow/1 and placed by path, unit, kind, lines and variable; each finding says whether it is judged or advisory (unused parameters always are). Report-only. |
| merge_suggestionsB | Clone-merge suggestions (ce.merge-report schema): every clone group — T1/T2 families, T3 pairs — anti-unified by the core's merge/1: the parameters a merged function takes with each member's text at them, the member kept, the lines saved, feasible or why not. Advisory. |
| architectureB | The architecture of the tree (ce.arch-report schema): directory layers and the arcs to cut out of their cycles (each with the file references behind it), file clusters and the files outside their cluster's directory, per-directory fan-in / fan-out / instability, and the files a change to |
| update_checkA | Whether a newer CodeEraser release exists (ce.update-report schema): the latest tag, that tag's committed SHA256 pins, and the one action for this install. Reads the release index over the network; places nothing — applying is a human act at the CLI or the GUI. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 21 tools
The set clusters into overlapping families: clone/check_duplication/docdup/join/merge_suggestions all concern duplication, and deadcode/flow/graph_sites all concern liveness, while scan/structure/architecture/trend/check all emit report-style metrics. The dense descriptions do differentiate them (T1/T2 vs T3, liveness verdicts vs in-function flow), but an agent could easily misselect within these families.
Names are uniformly lowercase, using snake_case for compounds (check_duplication, graph_sites, erase_log, merge_suggestions) and single words elsewhere (scan, erase, churn, flow, query). No camelCase or mixed verb styles, though the single-word vs compound mix means the pattern is readable rather than strictly uniform.
21 tools is at the heavy end but the domain (multi-faceted static analysis: duplication, dead code, structure, rules, query, diagnostics, updates) genuinely spans that many distinct reports. Each tool appears to earn its place rather than being redundant filler.
The surface covers the full analysis lifecycle: scanning, dedup, dead code, flow, structure, architecture, rules, Datalog query, trend, doctor, update_check, and erase audit. Applying an erase is deliberately a human act, and query/prelude fill the ad-hoc retrieval gap, so only minor workflow conveniences seem absent.