Tools

Everything Eclipse Beacon MCP exposes to your AI client: tools, resources and prompts.

5 capabilities5 tool

Automated heuristic

Tool docs quality: C — 5 tools graded

Average 60/100 for how clearly each description tells an AI client what the tool does, what it needs, and what it changes. It is not a security audit.

Grade A: 0
Grade B: 2
Grade C: 2
Grade D: 1
Grade F: 0

How grades work

gap_reporttool
Read-only analysis: assets with no covering policy, duplicate coverage on the same asset, premium spikes vs the prior term (default flag >15%). Findings are signals with evidence — never verdicts, never advice.
Description quality grade C, 55/100 · 3 to fix
- Has a description: passed20/20
  The tool has a description an AI client can read.
- Sensible length: passed20/20
  Length is in a useful range.
- Starts with what it does: not passed0/15
  Open with an action verb, for example “Search issues…” or “Creates a label…”.
- Mentions its inputs: not passed0/20
  Say what the tool needs, for example “by issue number” or “takes a file path”.
- Explains side effects: passed15/15
  Marked read-only.
- Gives usage guidance: not passed0/10
  Add when to use it, what it requires, or what to use instead.
policy_gettool
One policy in full, provenance included.
Description quality grade D, 45/100 · 4 to fix
- Has a description: passed20/20
  The tool has a description an AI client can read.
- Sensible length: not passed10/20
  A bit terse. One more clause about inputs or results would help.
- Starts with what it does: not passed0/15
  Open with an action verb, for example “Search issues…” or “Creates a label…”.
- Mentions its inputs: not passed0/20
  Say what the tool needs, for example “by issue number” or “takes a file path”.
- Explains side effects: passed15/15
  Looks read-only, so no disclosure is required. Saying “read-only” still helps.
- Gives usage guidance: not passed0/10
  Add when to use it, what it requires, or what to use instead.
policy_ingesttool
Parse a declarations-page-style document (JSON, or simple text dec page) from a local file. Returns a draft + diff with per-field provenance (source file, locator, confidence). Propose-only by default — nothing is written unless confirm=tru
Description quality grade B, 70/100 · 2 to fix
- Has a description: passed20/20
  The tool has a description an AI client can read.
- Sensible length: passed20/20
  Length is in a useful range.
- Starts with what it does: passed15/15
  Opens with an action verb.
- Mentions its inputs: not passed0/20
  Say what the tool needs, for example “by issue number” or “takes a file path”.
- Explains side effects: passed15/15
  Looks read-only, so no disclosure is required. Saying “read-only” still helps.
- Gives usage guidance: not passed0/10
  Add when to use it, what it requires, or what to use instead.
policy_listtool
Every policy in the graph with key dates, plus declared assets/persons and current alert windows.
Description quality grade B, 75/100 · 2 to fix
- Has a description: passed20/20
  The tool has a description an AI client can read.
- Sensible length: passed20/20
  Length is in a useful range.
- Starts with what it does: not passed0/15
  Open with an action verb, for example “Search issues…” or “Creates a label…”.
- Mentions its inputs: passed20/20
  The description mentions what the tool takes.
- Explains side effects: passed15/15
  Looks read-only, so no disclosure is required. Saying “read-only” still helps.
- Gives usage guidance: not passed0/10
  Add when to use it, what it requires, or what to use instead.
radar_reporttool
Ranked upcoming events: renewal (≤60d), premium due (≤14d), grace-period end, escrow/lender proof deadline, and static enrollment calendar entries (ACA / Medicare — dates move yearly, flagged [verify] on the surface). Each event carries sev
Description quality grade C, 55/100 · 3 to fix
- Has a description: passed20/20
  The tool has a description an AI client can read.
- Sensible length: passed20/20
  Length is in a useful range.
- Starts with what it does: not passed0/15
  Open with an action verb, for example “Search issues…” or “Creates a label…”.
- Mentions its inputs: not passed0/20
  Say what the tool needs, for example “by issue number” or “takes a file path”.
- Explains side effects: passed15/15
  Looks read-only, so no disclosure is required. Saying “read-only” still helps.
- Gives usage guidance: not passed0/10
  Add when to use it, what it requires, or what to use instead.