Collide MCP

remember

Destructive

Save a durable team fact: a decision, constraint or gotcha, not code state. Anchor facts about specific code (anchor="auth.py::validate_token", "auth.py" or "src/auth/"): anchored notes reach whoever touches that code and are marked stale when it is rewritten. When the human states a standing decision ("remove X", "never Y", "only Z"), save it with kind="decision" anchored to the code it covers: it never goes stale, and each agent's first write there is held once with it. Writes the note(s) and returns their ids.

Input Schema

TableJSON Schema

Name	Required	Description
`fact`	No	The note, one or two sentences.
`kind`	No	note (default), or decision: the owner's standing decision; needs an anchor.
`tags`	No	Labels to find it by with recall.
`facts`	No	Several notes at once: [{fact, anchor?, tags?, kind?}]. Decision items with the same fact are one decision over all their anchors.
`anchor`	No	Where it applies: path::symbol, a file, or dir/.
`repo_id`	No	The repository id, e.g. github.com/org/repo (the Collide briefing names it). Omit it to use the repo your session works in.
`supersedes`	No	Id of a note or decision this one replaces; the old one is retired. kind="decision" with no fact withdraws that decision.

Schema Changelog

Changes observed during successful MCP inspections.

Changed5 schema fields changedOct 18, 2026

changedInput schema / properties / facts / description

Previous value: -"Several notes at once: [{fact, anchor?, tags?}]."New value: +"Several notes at once: [{fact, anchor?, tags?, kind?}]. Decision items with the same fact are one decision over all their anchors."

addedInput schema / properties / kind

Added value: +{
+  "description": "note (default), or decision: the owner's standing decision; needs an anchor.",
+  "enum": [
+    "note",
+    "decision"
+  ],
+  "type": "string"
+}

changedInput schema / properties / repo_id / description

Previous value: -"The repository id, e.g. github.com/org/repo (repo_id in .collide/config.json)."New value: +"The repository id, e.g. github.com/org/repo (the Collide briefing names it). Omit it to use the repo your session works in."

changedInput schema / properties / supersedes / description

Previous value: -"Id of a note this one replaces; the old note is retired."New value: +"Id of a note or decision this one replaces; the old one is retired. kind=\"decision\" with no fact withdraws that decision."

changedInput schema / required

Previous value: -[
-  "repo_id"
-]New value: +[]

Changed6 schema fields changedSep 27, 2026

addedInput schema / properties / anchor / description

Added value: +"Where it applies: path::symbol, a file, or dir/."

addedInput schema / properties / fact / description
```
Added value: +"The note, one or two sentences."
```

addedInput schema / properties / facts / description

Added value: +"Several notes at once: [{fact, anchor?, tags?}]."

addedInput schema / properties / repo_id / description

Added value: +"The repository id, e.g. github.com/org/repo (repo_id in .collide/config.json)."

addedInput schema / properties / supersedes / description

Added value: +"Id of a note this one replaces; the old note is retired."

addedInput schema / properties / tags / description

Added value: +"Labels to find it by with recall."

Changed2 schema fields changedSep 18, 2026

addedInput schema / properties / facts

Added value: +{
+  "items": {
+    "type": "object"
+  },
+  "type": "array"
+}

changedInput schema / required

Previous value: -[
-  "repo_id",
-  "fact"
-]New value: +[
+  "repo_id"
+]

Changed17 schema fields changedSep 18, 2026
- removedInput schema / properties / anchor / default
```
Removed value: -""
```
- removedInput schema / properties / anchor / description
```
Removed value: -"Pin this memory to specific code: 'path.py::symbol', 'path.py', or 'src/dir/'."
```
- removedInput schema / properties / anchor / title
```
Removed value: -"Anchor"
```
- removedInput schema / properties / fact / description
```
Removed value: -"The durable fact or decision to remember."
```
- removedInput schema / properties / fact / title
```
Removed value: -"Fact"
```
- removedInput schema / properties / repo_id / description
```
Removed value: -"Repository id — the git remote (e.g. github.com/acme/api) or a bare folder name; a #branch suffix scopes to that branch."
```
- removedInput schema / properties / repo_id / title
```
Removed value: -"Repo Id"
```
- removedInput schema / properties / supersedes / default
```
Removed value: -""
```
- removedInput schema / properties / supersedes / description
```
Removed value: -"The id of a prior memory this note replaces."
```
- removedInput schema / properties / supersedes / title
```
Removed value: -"Supersedes"
```
- removedInput schema / properties / tags / anyOf
```
Removed value: -[
-  {
-    "items": {
-      "type": "string"
-    },
-    "type": "array"
-  },
-  {
-    "type": "null"
-  }
-]
```
- removedInput schema / properties / tags / default
```
Removed value: -null
```
- removedInput schema / properties / tags / description
```
Removed value: -"Optional labels to categorize this memory."
```
- addedInput schema / properties / tags / items
```
Added value: +{
+  "type": "string"
+}
```
- removedInput schema / properties / tags / title
```
Removed value: -"Tags"
```
- addedInput schema / properties / tags / type
```
Added value: +"array"
```
- removedInput schema / title
```
Removed value: -"rememberArguments"
```
First observedSep 17, 2026

TDQS

A4.5/5.0

Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Goes well beyond the annotations by disclosing non-obvious lifecycle behavior: anchored notes reach whoever touches that code, are marked stale when it is rewritten, decisions never go stale, and each agent's first write at a decision's location is held once with it. It also states the write returns ids. It does not explain what the destructiveHint=true annotation implies (retiring notes via supersedes is only documented in the schema), so this falls just short of a 5.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Three sentences, front-loaded with the core purpose and with the anchor and decision semantics following in logical order; every sentence carries information. It is dense with parenthetical examples, which slightly hurts scannability, but there is no filler.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 7-parameter, zero-required mutation tool with no output schema, the description covers the important semantics (what to save, anchoring, decision behavior, returned ids). It omits supersedes/withdrawal semantics and batch `facts` handling, which are only described in the schema, but nothing critical to a correct first call is missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the baseline is 3, but the description adds real meaning for anchor (path::symbol, file, or dir/, and who it reaches) and kind (decision requires an anchor and behaves differently). tags and supersedes are left to the schema, so it improves on the baseline without covering everything.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Save a durable team fact: a decision, constraint or gotcha') and immediately excludes the wrong content type ('not code state'), which separates it from code-reading siblings like explain_code, get_symbol, recall and recap. An agent knows without opening the schema that this is the write-side of team memory.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicit conditional guidance: anchor notes that concern specific code, and the specific trigger 'When the human states a standing decision ("remove X", "never Y", "only Z")' routed to kind="decision" with an anchor. It also names when-not-to-use ('not code state') and ties decisions to the code they cover, so selection between plain notes, anchored notes and decisions is fully determined.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All Collide MCP tools Inspect

Try in Browser

Context Every Agent Can ShareAgents share what they learn, hand off work, and keep it when you switch models. Start with 5 GB free.fulcradynamics.comAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources

Need Help?