Skip to main content
Glama

Recover a persisted checkpoint into a NEW task/session (starts a model request)

dsh_recover_task

Starts a new recovered task and session from a dsh_prepare_recovery plan, using checked goals, a bounded summary and fresh instructions, and returns an idempotent taskId.

Instructions

Starts a NEW task, private SDK runtime and session from a plan produced by dsh_prepare_recovery, using the checked original goal/constraints, a bounded checkpoint summary and the current instructions. The old task tree is never written and its stages are never appended. The recorded PID and recorded verification argv are never reused or executed: fresh caller-supplied verification settings are required when the source configured verification, and original protected pins must still match. A durable atomic claim returns the same new taskId with alreadyRecovered=true on retries, including after a bridge restart. A source recorded accepted/starting/running requires explicit allowUnsettledSource=true; that is a caller acknowledgment, not bridge proof of old-process liveness.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "taskId": "<taskId>",
      "workspace": "<workspace>",
      "planToken": "<planToken>",
      "instructions": "<instructions>"
    },
    "name": "dsh_recover_task"
  }
}

Input Schema

NameRequiredDescriptionDefault
taskIdYes
budgetMsNo
maxTokensNo
planTokenYesplanToken returned by dsh_prepare_recovery for this exact source and workspace.
workspaceYes
acceptanceNo
autoRepairNo
autoContinueNoBounded same-live-session continuation ONLY after a real provider max_tokens/length turn end. Omission is disabled; deliveryPreset complete enables maxAttempts=2, noProgressLimit=2, totalBudgetMs=3600000. Explicit fields win. Storage/preview truncation, cancel, permission/credential failures, a dead runtime, shutdown and verifier-integrity failures never trigger it.
instructionsYesWhat this recovered NEW stage must do now. The old conversation is not restored.
deliveryPresetNo
deliverySourceNoFresh explicit source scope for the recovered task. The source task's historic paths and patch baseline are never replayed.
reasoningEffortNo
executionProfileNoFresh execution profile. Omitted inherits the source profile when it had one, preserving the same execution semantics.
verificationCommandNoFresh argv the bridge runs itself after the recovered stage. Required when the source configured verification; the recorded argv is never executed.
allowUnsettledSourceNoExplicit acknowledgment that the source was recorded accepted/starting/running and the old stop status is unknown. It is not proof that the old process is dead; recovery still starts a NEW session.
verificationTimeoutMsNo
protectedVerificationFilesNoFresh pins for current verifier files. When the source pinned files, the same real paths and original hashes must be supplied; additional pins are allowed.

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv1.0.8

TDQS

A4.3/5.0
Behavior5/5

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

It goes well beyond the annotations by disclosing that the old task tree is never written, that the recorded PID and verification argv are never reused or executed, that protected pins must still match, and that a durable atomic claim returns the same new taskId with alreadyRecovered=true on retries including after a bridge restart. It also explicitly disclaims the meaning of allowUnsettledSource ('caller acknowledgment, not bridge proof of old-process liveness'), which prevents a false liveness assumption. The annotation set gives destructiveHint=false and openWorldHint=true but says nothing about any of these guarantees.

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 size is justified by the tool's complexity and the safety constraints are front-loaded in the opening sentence. Some sentences pack multiple clauses (the PID/argv/verification sentence), which makes them dense but each clause carries distinct behavioral 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 17-parameter, mutation-capable tool with no output schema and a real model request on invocation, the description covers the danger zones an agent could get wrong: source state gating, verification argv freshness, pin hashes, and retry semantics. What is missing is coverage of the tuning/preset parameters (deliveryPreset, autoRepair, budget fields), which the schema largely leaves bare.

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?

Schema coverage is only 47% across 17 parameters, so the description needs to compensate and it does so for the highest-risk ones: verificationCommand (fresh argv, recorded argv never executed), protectedVerificationFiles (original paths and hashes must be resupplied), and allowUnsettledSource. However, roughly half the parameters (taskId, workspace, budgetMs, maxTokens, acceptance, autoRepair, deliveryPreset, reasoningEffort, verificationTimeoutMs) get no meaning from the description, leaving a real gap below the baseline.

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 first sentence gives a specific verb plus the exact resources created ('Starts a NEW task, private SDK runtime and session') and names the sibling that produces the required plan, dsh_prepare_recovery. It also states what inputs are consumed (checked goal/constraints, bounded checkpoint summary, current instructions), which distinguishes it from dsh_continue_task or dsh_submit_task.

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 makes clear this is a recovery path that must follow a dsh_prepare_recovery plan and explains the two conditional gates: fresh verification settings are required when the source configured verification, and allowUnsettledSource=true is mandatory for sources recorded accepted/starting/running. It stops short of an explicit contrast with dsh_continue_task or dsh_get_task, so the agent must infer that those are the non-recovery alternatives.

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