buddie
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| BUDDIE_LOG | No | Hook run log; 0 turns it off. | ~/.cache/buddie/hook.jsonl |
| BUDDIE_FETCH | No | Set to 0 so the hook uses stored snapshots only instead of downloading cited pages. | on |
| BUDDIE_FINAL | No | Regex marking a final answer that must carry anchors at any event; with none, the last message at Stop is the final answer; empty turns the rule off. | final_mark in [mandate] of buddie.toml, else none |
| BUDDIE_GATES | No | class=level,... overrides the default levels of the gates. | [gates] of buddie.toml |
| BUDDIE_INTRO | No | The intro mode's count per buddie.toml; 0 keeps none. | ~/.cache/buddie/intro.json |
| BUDDIE_REPOS | No | dir:dir pairs of repositories used to check repo anchors. | git repos around the working directory |
| GITHUB_TOKEN | No | Alternative GitHub API token used for pr: and ci: anchors (public repos work without one, within rate limits). | |
| BUDDIE_GITHUB | No | Set to 0 to make no GitHub calls; pr: and ci: anchors become UNCHECKABLE. | on |
| BUDDIE_STORES | No | dir:dir extra snapshot stores the hook reads along with the session's own. | none |
| BUDDIE_LESSONS | No | Set to 0 for no lessons at SessionStart. | on |
| BUDDIE_NUMBERS | No | gap, fail or off: how the receipt judges a number missing from its source. | from the numbers level |
| VERBATIM_STORE | No | Snapshot store used to keep the bytes of every cited source. | ./.verbatim |
| BUDDIE_SHARED_LOG | No | Second log, one file per session; 0 turns it off. | /mnt/project-files/buddie/hook, only if that folder exists |
| BUDDIE_GITHUB_TOKEN | No | GitHub API token used for pr: and ci: anchors (public repos work without one, within rate limits). | |
| BUDDIE_JOURNAL_TEXT | No | Set to 1 so the journal also keeps promise sentences, choice and decision cards as text. | [journal] text of buddie.toml, else off |
| BUDDIE_CONTEXT_LIMIT | No | Context tokens past which a non-final reply is returned; 0 turns it off. | 200000 |
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
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| verify_report | Check a report before accepting or passing it on: every quote must be in the bytes of the source it links to (verbatim), every |
| snapA | Snapshot URLs (bytes by sha256) so quotes can be taken from and checked against them. Call it before quoting a page, then copy the words from source_text and link the same URL. |
| source_textB | Visible text of a snapshot (URL or sha256 prefix). With grep, only the matches with context. Copy quotes from here character for character, put them in quotation marks “…” or «…», and link the snapshot's URL right after the quote in the same sentence. |
| summaryC | When no subagent is left: the summary built by code from the hook's records (what each subagent did with its receipt outcome, who was lost, the next steps with auto/ask verdicts, open PRs and red CI) and |
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 4 tools
Each tool targets a different stage: snap creates snapshots, source_text retrieves quotable text, verify_report validates quotes/anchors, and summary reports subagent session outcomes. Their descriptions cross-reference each other, but the core actions are distinct enough to avoid misselection.
All names use lowercase snake_case and are readable, but the set mixes verb-led (verify_report), noun-phrase (source_text), bare verb (snap), and bare noun (summary) forms. There is no single predictable verb_noun pattern.
Four tools is a tight, well-scoped set for a snapshot/verification workflow. Each tool has a clear role and none appear redundant or missing for the stated purpose.
The set covers creating snapshots, extracting quotable text, verifying reports, and summarizing subagent sessions. It may lack explicit listing or searching of existing snapshots/reports, but the core lifecycle is present and workable.