CtrlK
BlogDocsLog inGet started
Tessl Logo
PluginPublic

ainativedev/aidevcon-2026-nyc

AI Native DevCon 2026 New York — practical agent skills grounded in conference talks

Quality

86%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

Overview
Quality
Evals
Security
Files

application-template.mdskills/talk-moss-concurrent-agent-risk/references/

Concurrent-agent work-item risk record

Use this worksheet for one workflow before increasing parallel execution. It is a Tessl adaptation, not the queue or budget implementation described by Moss.

Work identity and authority

Logical work item: Ticket and version: Repository and base revision: Implementation, review, and orchestration actors: Authorized actions and existing merge policy: External systems that record progress:

Draw or describe the transitions between requested, claimed, running, reviewable, completed, failed, and retrying states as appropriate. State which records are authoritative and where disagreement could occur.

Five failure probes

RiskConcrete probeExpected responseEvidence
Shared stateTwo separate tickets modify one generated history file
StalenessAnother valid feature lands before rebase
ConvergenceThe same concern returns in different wording
DuplicationTwo workers receive one event concurrently
BudgetAn agent comments through a human identity

Adapt the probes to the real system; do not claim they have run. For shared state, name the correct owner and whether work should be staged. For staleness, identify the old and new behavior that must both survive.

Recovery and stopping design

Work-item identifier: How ownership is claimed: What prevents duplicate external effects: How a crashed worker's partial output is reconciled: How repeated findings are recognized: What constitutes meaningful review progress: Cost sources counted across workers: Budget and elapsed-time limits: Stop, preserve-evidence, and continuation behavior:

Treat numeric limits as project choices. Record the reason for each and who can intentionally continue the work. Do not interpret an ordinary comment as unlimited fresh authorization.

Decision

Observed weakness: Proposed smallest control: Validation plan: Remaining uncertainty: Scope suitable for the next bounded trial:

Separate reported incidents, inspected local findings, and proposed controls. A clean merge or a green test alone does not show that another feature survived.

skills

talk-moss-concurrent-agent-risk

tile.json