cendor-contextkit

Treat the context window like a packed suitcase, not a string you concatenate. Declare blocks with priorities and eviction rules; contextkit fits them to a token budget and tells you exactly what it kept, shrank, and dropped.

Every assembled prompt comes with a receipt.

PyPI license · pip install cendor-contextkit

from cendor.contextkit import Context, Block

ctx = Context(budget_tokens=8000, model="claude-opus-4-8", reserve_output=1000, order="attention")
ctx.add(Block(system_prompt, priority=10, pin=True, role="system"))
ctx.add(Block(retrieved_docs, priority=5, evict="compress"))        # uses squeeze if installed
ctx.add(Block(messages=chat_history, priority=3, evict="drop_oldest"))  # peels OLDEST turns
ctx.add(Block(user_msg, priority=9, pin=True, role="user"))

messages = ctx.assemble()      # provider-ready, guaranteed within budget — including framing
print(ctx.report())            # the receipt: kept / truncated / dropped + token math
preview = ctx.whatif(budget_tokens=4000)   # same inputs, tighter budget, no commit

Highlights

  • Token-budgeted packing — declare Blocks with priority and pin; assemble() fits them into the budget deterministically. Pinned blocks are never evicted (raises BudgetError if they alone overflow).
  • Per-block eviction — drop_oldest · truncate (keep head/tail, with a …[truncated] marker) · summarize (sync, or async via aassemble()) · compress (via squeeze — reversible: the receipt's BlockDecision.handle.expand() restores the original, and compression is sized against your model) · or any custom EvictionStrategy.
  • Real chat-history — Block(messages=[…]) peels the oldest turns to fit (a sliding window) — never mangling a turn.
  • An honest receipt — report().used == tokens.count(assemble(), model): packing charges the per-message framing providers add, so a "full" prompt never quietly overflows once sent.
  • Attention-aware ordering — order="default" · "attention" (lost-in-the-middle) · "cache" (stable prefix for prompt-cache hits).
  • Provider adapters & multimodal — for_anthropic() / for_gemini() / for_bedrock(); per-image image_tokens (int or resolution-aware callable); whatif(budget) previews; use_compressor() swaps the compression backend.

Inbound — call it before the model call; report() flows onto core's bus, so acttrace records what the model actually saw.

See docs/contextkit.md · CHANGELOG. Part of the Cendor stack — github.com/cendorhq/cendor-libs. Powered by PowerAI Labs. Apache-2.0; provided "as is", without warranty — use at your own risk (LICENSE §7–8).

Project links

Data verified by PyPI on Jul 10, 2026
Data provided by the project maintainers, verified at the time the release was uploaded to PyPI.

Key dates

PyPI data
Data sourced directly from PyPI's database.
  • Released:
Newer release available (1.1.0)

1 maintainer

PyPI data
Data sourced directly from PyPI's database.
Avatar for powerailabs from gravatar.com powerailabs

Credits

Author: Raghav Mishra (PowerAI Labs)

GitHub Statistics

Data verified by PyPI on Jul 10, 2026
The GitHub source repository was provided by the project maintainers and verified by PyPI at the time of upload. Stars, forks, and open issues/PRs are derived from that repository and have not been independently verified.

License expression

Apache-2.0
View SPDX License List

Requires

Python >=3.11

Provides Extra

squeeze

Tags

agents ai context-window llm prompt rag token-budget

Classifiers

Report project as malware

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

cendor_contextkit-1.0.2.tar.gz (22.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

cendor_contextkit-1.0.2-py3-none-any.whl (17.6 kB view details)

Uploaded Python 3

File details

Details for the file cendor_contextkit-1.0.2.tar.gz.

File metadata

  • Download URL: cendor_contextkit-1.0.2.tar.gz
  • Upload date:
  • Size: 22.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.13

File hashes

Hashes for cendor_contextkit-1.0.2.tar.gz
Algorithm Hash digest
SHA256 a739d1496723e6f8686d5173fec0a283db06c5d34a6725ad993f6644e574a333
MD5 5bde4668efbd6fe519bdc0435074cd2a
BLAKE2b-256 522eaf8898f554d325fc06f118ef6888a364560a849eb98796ad35c6499acbf8

See more details on using hashes here.

Provenance

The following attestation bundles were made for cendor_contextkit-1.0.2.tar.gz:

Publisher: release.yml on cendorhq/cendor-libs

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file cendor_contextkit-1.0.2-py3-none-any.whl.

File metadata

File hashes

Hashes for cendor_contextkit-1.0.2-py3-none-any.whl
Algorithm Hash digest
SHA256 b5ec73ea35ffe1f3a4803d32e7db0dabc96ee86a7af8d8bb0da07c63720c4919
MD5 9b02276e7985173936c735ae662e893a
BLAKE2b-256 c690b10cbd095aafc9f375b2733ac9b7fb896855dcab0fb403a911f66f853e4a

See more details on using hashes here.

Provenance

The following attestation bundles were made for cendor_contextkit-1.0.2-py3-none-any.whl:

Publisher: release.yml on cendorhq/cendor-libs

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Release history Release notifications | RSS feed

1.1.0

2 files

1.0.3

2 files

This release

1.0.2 This release

2 files

1.0.1

2 files

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page