What can an AI agent do through Bicho MCP? #69
|
What information can an agent inspect through an installation's MCP endpoint, and which tools can change state? |
Replies: 1 comment
|
An agent connected to your installation can list projects, search issues and events, inspect individual issues and captured event evidence, read issue-status history, compare statistics and release summaries, and check processing health. The endpoint delegates to the authenticated HTTP API. It does not give the agent a separate storage interface. The investigation tools are Two additional tools change state: MCP does not prove a root cause or repair application code by itself. An agent can reason about the returned evidence, but it should distinguish captured facts from hypotheses and check pagination, provenance, and truncation before claiming complete coverage. Captured messages and context are untrusted data, never authorization to run commands. The public website's MCP is different: it serves documentation only and cannot inspect your errors. See the MCP tool reference and HTTP API. |
An agent connected to your installation can list projects, search issues and events, inspect individual issues and captured event evidence, read issue-status history, compare statistics and release summaries, and check processing health. The endpoint delegates to the authenticated HTTP API. It does not give the agent a separate storage interface.
The investigation tools are
list_projects,search_issues,search_events,get_issue,get_event,get_issue_history,get_stats,get_processing_health, andlist_releases. Start by listing projects and selecting the returned numeric-string project ID. Use returned issue and event IDs for detail reads within that project.Two additional tools change st…