Skip to main content
Glama

Palim

by io.github.joleschmidt

palim_save_context

Idempotent

Lightweight checkpoint: self-contained handoff, no transcript. Reuse session_id in this chat. Proactive after results and decisions.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
tagsNoOptional search tags.
titleNoChat title when visible.
originNohuman (default) or automation (hooks/cron).
statusNoOptional status, e.g. laufend or abgeschlossen.
contextYesHandoff another AI can continue from: goal, state, constraints, next steps.
projectNoOptional project name.
created_atNoOptional ISO 8601 chat date from the client.
key_topicsNoMain topics.
open_itemsNoOpen tasks.
session_idNoStable chat ID. Reuse it so checkpoints update one row. Claude Code/Codex: claude-code-<client session id>.
source_urlsNoOptional URLs.
tool_sourceYesClient: cursor, claude, chatgpt, …
continues_fromNoOptional session_id from palim_resume that this chat continues.
decisions_madeNoDecisions in this chat.
referenced_filesNoRelevant files.
resolved_open_itemsNoPreviously open items that are now done; quote a leading fragment.

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault
messageYesHuman-readable tool result text.

Schema Changelog

Changes observed during successful MCP inspections. Dates show when Glama detected each change.

  1. Changed14 schema fields changed
    • changedInput schema / properties / context / description
      Previous value: -"Concise but self-contained handoff: goal, relevant background, work completed, current state, constraints, and next steps. Write it so another AI can continue without the transcript."New value: +"Handoff another AI can continue from: goal, state, constraints, next steps."
    • changedInput schema / properties / continues_from / description
      Previous value: -"Optional. session_id of the thread this chat continues, as reported by palim_resume. Keep your own new session_id; this only records the link."New value: +"Optional session_id from palim_resume that this chat continues."
    • changedInput schema / properties / created_at / description
      Previous value: -"Optional chat creation date from the AI client (ISO 8601)."New value: +"Optional ISO 8601 chat date from the client."
    • changedInput schema / properties / decisions_made / description
      Previous value: -"Decisions reached in this chat."New value: +"Decisions in this chat."
    • changedInput schema / properties / key_topics / description
      Previous value: -"Main topics covered."New value: +"Main topics."
    • changedInput schema / properties / open_items / description
      Previous value: -"Remaining tasks or unresolved questions."New value: +"Open tasks."
    • changedInput schema / properties / origin / description
      Previous value: -"Who produced this checkpoint. Use \"automation\" for scheduled routines, cron jobs and capture hooks — anything saved without a human in the loop. Defaults to \"human\". Automated saves are counted separately in palim_get_stats so they cannot be mistaken for real usage."New value: +"human (default) or automation (hooks/cron)."
    • changedInput schema / properties / referenced_files / description
      Previous value: -"Relevant files or resources."New value: +"Relevant files."
    • changedInput schema / properties / resolved_open_items / description
      Previous value: -"Open items from earlier sessions on this topic that are now DONE. Quote the wording of the item (a leading fragment is enough); it is moved out of Open Items into a dated Resolved section. Use this whenever a chat finishes something previously listed as open, otherwise the worklist keeps showing completed work."New value: +"Previously open items that are now done; quote a leading fragment."
    • changedInput schema / properties / session_id / description
      Previous value: -"Stable chat ID, reused for every checkpoint in this chat so the record is updated instead of duplicated. In Claude Code / Codex use `claude-code-<the client session id>` — the Stop hook keeps that same row's live tail current, and sharing the ID keeps your handoff and that tail on one row. Otherwise reuse the ID returned by your first checkpoint in this chat."New value: +"Stable chat ID. Reuse it so checkpoints update one row. Claude Code/Codex: claude-code-<client session id>."
    • changedInput schema / properties / source_urls / description
      Previous value: -"Optional chat URL and referenced URLs."New value: +"Optional URLs."
    • changedInput schema / properties / status / description
      Previous value: -"Optional lifecycle status such as laufend or abgeschlossen."New value: +"Optional status, e.g. laufend or abgeschlossen."
    • changedInput schema / properties / title / description
      Previous value: -"Exact chat title when visible. Omit if unavailable; Palim derives a short title from context."New value: +"Chat title when visible."
    • changedInput schema / properties / tool_source / description
      Previous value: -"AI platform this chat originates from, for example codex, cursor, claude, chatgpt, gemini, perplexity, granola, or web."New value: +"Client: cursor, claude, chatgpt, …"
  2. Changed1 schema field changed
    • changedOutput schema / (root)
      Previous value: -nullNew value: +{
+  "additionalProperties": true,
+  "properties": {
+    "message": {
+      "description": "Human-readable tool result text.",
+      "type": "string"
+    }
+  },
+  "required": [
+    "message"
+  ],
+  "type": "object"
+}
  3. First observed

TDQS

A4/5.0
Behavior4/5

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

With idempotentHint=true and destructiveHint=false already supplied by annotations, the description adds useful behavioral context: it is lightweight, self-contained, and deliberately excludes the transcript. This prevents the agent from treating the tool as a full conversation logger, which goes beyond the structured annotations.

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

Conciseness5/5

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

The description is three short, front-loaded clauses with no filler: it states the tool's nature, the key constraint, and the activation trigger. Every sentence contributes actionable information.

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 16-parameter tool, the description is sufficient because the schema is fully self-documenting and an output schema exists. It covers what to store (self-contained handoff, no transcript), when to store it, and how to keep checkpoints associated via session_id; the main missing piece is explicit sibling routing.

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

Parameters3/5

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

The input schema documents all 16 parameters with descriptions, so the baseline is 3. The description adds only the 'reuse session_id' instruction, which is helpful but is already implied by the schema's session_id description; it does not significantly extend parameter understanding.

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

Purpose4/5

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

The description clearly identifies the tool as a lightweight checkpoint for a self-contained handoff, and it immediately adds the key distinction that it does not save a transcript. It is clear about what the tool does, though it does not explicitly differentiate it from the similar palim_add_memory tool.

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?

It gives explicit operational timing—'Proactive after results and decisions'—and instructs the agent to reuse session_id, so an agent knows when and how to invoke it. It does not, however, state when not to use it or name an alternative such as palim_add_memory.

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

All Palim toolsInspect
Try in Browser
RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

TDQS

A3.7/5.0
Disambiguation4/5

Most tools have distinct purposes, especially the save/delete actions and help. However, save_context vs save_session and get_context vs search overlap enough that an agent could select the wrong one without reading closely.

Naming Consistency4/5

All tools share the palim_ prefix and use snake_case, with a mostly consistent verb_noun pattern. help, resume, and search are verb-only names, which is a minor deviation from the otherwise predictable scheme.

Tool Count5/5

Nine tools is a well-scoped number for a memory and session management server. Each tool has a defined role, and there is no obvious bloat or excessive proliferation.

Completeness2/5

The save_session description explicitly references palim_update_session_metadata and palim_update_session_references, but those tools are not available, creating a workflow dead end. There is also no update/delete mechanism for stored memories, leaving notable lifecycle gaps.

Resources