Skip to main content
Glama

Add workfile memory

project_memory_add

Adds a typed memory record—learning, decision, incident, convention, or context—to project memory, capturing knowledge that should alter future behavior.

Instructions

Create a typed learning, decision, incident, convention or temporary context record. For knowledge that should change future behaviour — not for narrating what happened this session.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
bodyNoMarkdown body, written below the frontmatter.
tagsNoFree-form tags for filtering.
scopeNoSubject matter this applies to — topics, not file paths.
titleYesThe knowledge itself, stated as a claim rather than a topic.
ownersNoWho is answerable for keeping this accurate.
statusNoLifecycle status. Each collection has its own vocabulary — a decision is proposed or accepted, an incident is open or resolved. Project-declared, so the accepted values vary; project_workspace reports them.
expiresNoFor context: the date after which this stops being true, YYYY-MM-DD.
relatedNoIDs of records worth reading alongside this one.
categoryNoOptional sub-classification within the collection.
decidersNoFor decisions: who made the call.
severityNoFor incidents: how bad it was.
collectionYesWhich kind of memory this is. learnings: something discovered that changes how the next attempt goes. decisions: a choice made, with its rationale (ADR). incidents: something that broke, and what it cost. conventions: a rule the project now follows. context: temporary situational state, which expires.
confidenceNoHow well established this is — for learnings, whether it has been seen once or many times.
started_atNoFor incidents: when it began, YYYY-MM-DD.
supersedesNoIDs of memory records this replaces. Prefer project_memory_supersede, which updates both sides.
occurrencesNoHow many times this has been observed.
resolved_atNoFor incidents: when it was resolved, YYYY-MM-DD.
graduated_toNoIDs of the durable records a learning has been promoted into.
review_afterNoDate after which this should be re-read, YYYY-MM-DD.
corrective_actionsNoFor incidents: IDs of the cards that stop it recurring.

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault
recordYesA canonical record with its body, revision and reference graph.
Behavior3/5

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

Annotations declare readOnlyHint=false, so mutation is implied by the annotation already. The description adds the intended-use context (behavior-changing knowledge vs session narration). However, it doesn't disclose what happens on record creation (return value, whether duplicates are checked, side effects like index updates, or how errors are surfaced). With 4 annotation fields all set to defaults/false, the description doesn't compensate for the lack of behavioral detail.

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?

The description is two sentences and front-loads the purpose immediately in sentence one. Sentence two provides the essential exclusion criterion. It's efficient with no filler. Slightly more guidance on when NOT to use (which alternative tool for narration) would push it to 5, but as written it earns its place.

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 20-parameter tool, the description relies heavily on the schema (100% coverage) and output schema being present to explain return values. The description correctly focuses on the selection decision (which collection type, purpose vs narration). The phrase about status vocabulary varying by collection and being project-declared is valuable cross-tool guidance referencing project_workspace. The main gap is no explicit statement about whether related records or supersede chains are validated/created atomically.

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 description need not repeat parameter meaning. However, the description adds meaningful cross-collection guidance, particularly the collection enum values ('learnings: something discovered... decisions: a choice made... incidents: something that broke...') in the schema itself, and pointers like 'prefer project_memory_supersede' for supersedes, and 'project_workspace reports them' for status. This enriches param semantics beyond bare types.

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?

The description clearly states the verb (Create) and resource (typed learning, decision, incident, convention or temporary context record). It distinguishes from sibling tools (project_memory_list, project_memory_patch, project_memory_supersede, project_memory_graduate) by being the add operation. The phrase 'not for narrating what happened this session' adds a clear exclusion criterion.

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

Usage Guidelines4/5

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

The description explicitly states the tool is for 'knowledge that should change future behaviour' and excludes 'narrating what happened this session'. While it doesn't name specific alternative tools for the 'narrating' case, the collection parameter enum and cross-references to project_memory_supersede and project_memory_graduate in schema properties provide contextual alternatives. It could more explicitly say which sibling handles narration.

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

Install Server

Other Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/illodev/workfile'

If you have feedback or need assistance with the MCP directory API, please join our Discord server