context-shredding

Runtime context compression for agents approaching model context limits. Defines 3 compression tiers (full/summarized/minimal) with per-artifact templates. USE FOR: reducing artifact loading size at runtime, context budget management. DO NOT USE FOR: diagnostic context auditing (use context-optimizer), Azure infrastructure.

Quality

71%

Does it follow best practices?

Impact

Pending

No eval scenarios have been run

Securityby

Passed

No known issues

Optimize this skill with Tessl

npx tessl skill review --optimize ./.github/skills/context-shredding/SKILL.md

Quality

Discovery

85%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

This is a strong description that clearly defines its niche with specific capabilities, explicit use/non-use triggers, and anti-conflict guidance. The main weakness is that trigger terms lean technical rather than matching natural user language. The DO NOT USE FOR clause is an excellent practice for reducing skill selection conflicts.

Suggestions

Add more natural-language trigger terms users might say, such as 'running out of context', 'context window full', 'token limit', or 'reduce context size'.

Dimension	Reasoning	Score
Specificity	Lists specific concrete actions: runtime context compression, 3 compression tiers (full/summarized/minimal), per-artifact templates, reducing artifact loading size, context budget management. These are concrete, well-defined capabilities.	3 / 3
Completeness	Clearly answers both what (runtime context compression with 3 tiers and per-artifact templates) and when (USE FOR / DO NOT USE FOR clauses explicitly define triggers and anti-triggers, including distinguishing from context-optimizer and Azure infrastructure).	3 / 3
Trigger Term Quality	Includes some relevant terms like 'context compression', 'context limits', 'compression tiers', 'artifact loading size', 'context budget', but these are fairly technical. Missing more natural user phrases like 'running out of context', 'too much context', 'context window full', or 'token limit'.	2 / 3
Distinctiveness Conflict Risk	Explicitly distinguishes itself from related skills by including 'DO NOT USE FOR' clauses naming context-optimizer for diagnostic auditing and Azure infrastructure. The specific niche of runtime compression tiers is well-defined and unlikely to conflict.	3 / 3
	Total	11 / 12 Passed

Implementation

57%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

This skill is well-structured with clear progressive disclosure and a logical tier system, but suffers from duplicated threshold information and lacks concrete executable examples for context estimation. The workflow would benefit from a validation step to confirm the selected tier is appropriate after loading.

Suggestions

Remove the duplicated tier thresholds — keep either the table or the Tier Selection Protocol block, not both, to improve conciseness.

Add a concrete, executable example of context estimation (e.g., a Python snippet or a specific heuristic like 'count messages × avg tokens') to make the 'estimate context usage' step actionable.

Add a validation checkpoint after loading with compression, e.g., 'Verify the loaded content contains the key decisions needed for the current task; if critical info is missing, try the next higher tier.'

Dimension	Reasoning	Score
Conciseness	Mostly efficient but has some redundancy — the tier thresholds are repeated in both the table and the 'Tier Selection Protocol' text block. The model-specific limits (Opus, GPT-5.3-Codex) add useful detail but the overall content could be tightened by eliminating the duplicated threshold information.	2 / 3
Actionability	Provides a clear protocol and thresholds, but the guidance is procedural rather than executable — there's no concrete code or command to estimate token usage, and 'count approximate conversation tokens' is vague. The compression templates themselves are deferred to a reference file rather than shown inline even briefly.	2 / 3
Workflow Clarity	The multi-step process is clearly sequenced (estimate → select tier → apply template → prioritize older artifacts), but there are no validation or feedback steps. If the wrong tier is selected or compression fails, there's no guidance on how to detect or recover from that. For a process that could degrade agent performance, a verification checkpoint is warranted.	2 / 3
Progressive Disclosure	Clean structure with a concise overview in the skill file and a single well-signaled reference to compression-templates.md for detailed per-artifact templates. The reference index table is clear and one level deep. The skill variant table also helps with navigation.	3 / 3
	Total	9 / 12 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 11 / 11 Passed

Validation for skill structure

No warnings or errors.

Repository: jonathan-vella/azure-agentic-infraops
Commit: ec7b8ff

Reviewed: 6 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.