Codna
OfficialServer Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| GITHUB_TOKEN | No | GitHub token used by codna_fix with open_pr=true and codna_report_bug to interact with GitHub. |
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 | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| codna_triageA | Understand a repository and locate the code relevant to an issue. The sibling tools act, this one maps: codna_fix plans/acts on one bug and codna_recall retrieves stored code memory — triage is the zero-token map + suspect-file list you run first. Requires the one-time free repo: a local path or a git URL (default: current directory). issue: optional description of what you're looking for. Returns suspect files + the context-reduction the engine achieved (0 LLM tokens). |
| codna_fixA | Find and fix a bug. Runs the full Codna agent + engine + risk simulation. Requires the one-time free Returns the root cause, confidence, model, and (when open_pr) the pull_request_url. Failures come back as 'codna_fix error: ...' text — the tool never raises. |
| codna_secureA | Prove which scanner findings are reachable. Ingests a SARIF 2.1.0 report (CodeQL / Semgrep / Snyk / Trivy), classifies each finding (exploitable / production-reachable / unreachable / unknown) via the local engine, and reports which are autofix-eligible. Read-only, deterministic, 0 LLM tokens — the SARIF file is the only input read; the repo is never scanned and the agent sidecar never starts.
Requires the one-time free |
| codna_recallA | Recall relevant code from the local on-device Telys memory — semantic + lexical search over the symbols this machine has already indexed, with zero LLM tokens and zero network calls. Use when you need the code behind a concept ("where is SARIF provenance validated?")
rather than a whole-repo map (codna_triage) — and when you want it offline. The index
is built on first use (an empty index auto-indexes the repo, once) and is stored
outside the git checkout. On a fresh machine, first use requires the one-time free
repo: a local path or git URL (default: current directory, or the server default set
by |
| codna_report_bugA | File a report to thyn-ai/feedback — the public front door for algenta, codna, telys,
and sqai. Files the same shape of issue Requires the one-time free title: a short summary, e.g. "codna fix hangs on a monorepo".
body: what happened, or what you want — as much detail as you have.
product: one of algenta, codna, telys, sqai, accounts, docs (default: codna).
include_diagnostics: attach the same redacted output |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| fix_bug | Templated prompt: ask codna to locate then fix a bug. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| capabilities | codna's MCP capabilities — @-mentionable so the client knows what codna offers. |
TDQS
Scored across 5 tools
Each tool targets a clearly distinct workflow: triage maps issues to suspect files, fix performs the actual bug fix, recall searches local code memory, secure classifies SARIF findings, and report_bug files feedback. There is minor conceptual overlap between triage and recall, but their descriptions draw a clear line between whole-repo mapping and concept-level symbol recall.
All tools share the codna_ prefix and use lowercase imperative verbs, giving a predictable and recognizable pattern. The only small deviation is report_bug, which appends an object noun, while the others are bare verbs.
Five tools is well-scoped for a coding-agent server: triage, fix, recall, secure, and bug reporting each cover a distinct high-level capability without redundancy. The count feels intentional rather than padded.
The toolset covers the main workflows an agent needs: locating relevant code, fixing bugs, recalling indexed symbols, verifying security findings, and reporting issues. Minor gaps exist such as no general issue-creation tool or explicit test/verify step, but these can be worked around via the CLI or GitHub.